A multi-dimensional approach to disinformation Report of the independent High level Group on fake news and online disinformation Communications Communications Networks, Content Content Networks, and Technology Technology and Manuscript completed in March 2018 This document has been prepared for the European Commission however it reflects the views only of the High Level Group members, and the Commission cannot be held responsible for any use which may be made of the information contained therein. Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use that might be made of the following information. Cover photo © iStock.com/chombosan Luxembourg: Publications Office of the European Union, 2018 print PDF ISBN 978-92-79-80419-9 ISBN 978-92-79-80420-5 doi:10.2759/0156 doi:10.2759/739290 KK-01-18-221-EN-C KK-01-18-221-EN-N © European Union, 2018 Reuse is authorised provided the source is acknowledged. The reuse policy of European Commission documents is regulated by Decision 2011/833/EU (OJ L 330, 14.12.2011, p. 39). For any use or reproduction of photos or other material that is not under the EU copyright, permission must be sought directly from the copyright holders. Printed in Belgium Printed on white chlorine-free paper EUROPEAN COMMISSION A multi-dimensional approach to disinformation Report of the independent High level Group on fake news and online disinformation 2018 Directorate-General for Communication Networks, Content and Technology 2 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION Foreword by Mariya Gabriel, Commissioner for Digital Economy and Society I am delighted to preface the final report of the Commission High Level Expert Group on Fake News and Online Disinformation. Upon set-up of the group early this year, I tasked the HLEG to put forward possible options to counter disinformation spread online and to help develop a comprehensive EU strategy for tackling disinformation. I requested its members to work on the definition of the phenomenon, to identify measures already taken by various stakeholders and to establish key principles and general objectives for the short and long term. When tackling disinformation online, four principles underlie my endeavour: transparency so that citizens have clear information about news sources and funding; diversity of information both online and offline because this fuels critical judgement; credibility of information must be obvious to citizens; and inclusivity as there can be no long term, sustainable solution without commitment from all the parties involved. I am pleased to see that these principles are well reflected in the recommendations made by the members of the HLEG in their final report. Given the complexity of the problem, which requires a multistakeholder solution, there is no single lever to achieve these ambitions and eradicate disinformation from the media ecosystem. Improving the ability of platforms and media to address the phenomenon requires a holistic approach, the identification of areas where changes are required, and the development of specific recommendations in these areas. Thanks to the High Level Group and to its chairman Professor Madeleine de Cock Buning for the great work within tight deadlines, we now have at our disposal a wide array of material that will help us put forward a number of policy initiatives to better address the risks posed by disinformation spread online. My intention is to trigger a free, pluralistic democratic, societal, and economic debate in Europe. Fully respecting fundamental EU values, e.g. freedom of speech, media pluralism and media freedom, the recommendations of the HLEG put forward ways to ensure better media and information literacy in Europe, a diverse and sustainable news media ecosystem, fostered transparency, algorithm accountability and trust-enhancing practices. Last but not least, it seeks to further empower users and journalists. This report is just the beginning of the process and will feed the Commission reflection on a response to the phenomenon. Our challenge will now lie in delivering concrete options that will safeguard EU values and benefit every European citizen. Foreword Foreword by Professor Madeleine de Cock Buning In January 2018 Commissioner Mariya Gabriel convened the High Level Expert Group (HLEG) to advise the European Commission on fake news and disinformation online, Broadly, the HLEG was tasked with defining the scope of the phenomenon, identifying the roles and responsibilities of relevant stakeholders, and formulating recommendations. I am very proud to present our Report. It represents the first time that members from the most relevant stakeholder groups have agreed upon steps stakeholders can jointly take to provide European citizens with more trustworthy information and make them better equipped to navigate the online environment. Taking freedom of expression and the right to receive and impart information as our starting point, we managed to fashion a widely-supported, multi-dimensional and concrete approach to the complex problem of disinformation. In this Report, we favour the word “disinformation” over “fake news.” Disinformation, as used in the Report, includes all forms of false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit. Our main message is that the best responses to disinformation are multi-dimensional, with stakeholders collaborating in a manner that protects and promotes freedom of expression, media freedom, and media pluralism. Another key message is that effective action will require continuous research on the impact of disinformation, increased transparency, and access to relevant data, combined with evaluation of responses on a regular, ongoing basis. This is particularly important as disinformation is a multi- faceted and evolving problem that does not have one single root cause. The recommendations in this Report aim to provide shortterm responses to the most pressing problems, longer-term responses to increase societal resilience to disinformation, and a framework for ensuring that the effectiveness of these responses is continuously evaluated and, new evidencebased responses are developed. Our recommendations are addressed to the Commission and the Member States as well as to stakeholders: platforms, news media and civil society organisations. We believe that our Report provides the ingredients for an effective pan-European strategy to tackle the complex and disruptive problem of disinformation. I thank Commissioner Gabriel for entrusting the HLEG with such an important task. I also thank the Directorate-General for Communication Networks, Content and Technology for their excellent support as secretariat to the HLEG. Lastly, I wholeheartedly thank all of my colleagues on the HLEG for their engagement in and dedication to this collective endeavour. We were under the pressure of time constraints and had many spirited discussions. Together, we sharpened our views and, as a result, were able to deliver what I believe is a firm and powerful Report that provides solid analysis and advice. I feel it is indeed a great achievement that the members of the group were able to look to the collective importance of well-informed European citizens even if that sometimes meant setting aside their own predilections. 3 EXECUTIVE SUMMARY Executive summary In January 2018, the European Commission set up a highlevel group of experts («the HLEG») to advise on policy initiatives to counter fake news and disinformation spread online. The HLEG consisted of 39 members and was chaired by Prof. Dr. Madeleine de Cock Buning. Its members had different backgrounds, including academia and journalism, written press and broadcasting organizations, online platforms as well as civil society and fact-checking organizations. The HLEG’s tasks were to advise the Commission on all issues arising in the context of false information spread across traditional and social media and on possible ways to cope with its social and political consequences. The main deliverable of the HLEG was a report designed to review best practices in the light of fundamental principles, and suitable responses stemming from such principles. The analysis presented in this Report starts from a shared understanding of disinformation as a phenomenon that goes well beyond the term «fake news». This term has been appropriated and used misleadingly by powerful actors to dismiss coverage that is simply found disagreeable. Disinformation as defined in this Report includes all forms of false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit. It does not cover issues arising from the creation and dissemination online of illegal content (notably defamation, hate speech, incitement to violence), which are subject to regulatory remedies under EU or national laws. Nor does it cover other forms of deliberate but not misleading distortions of facts such a satire and parody. Problems of disinformation are deeply intertwined with the development of digital media. They are driven by actors — state or non-state political actors, for-profit actors, media, citizens, individually or in groups — and by manipulative uses of communication infrastructures that have been harnessed to produce, circulate and amplify disinformation on a larger scale than previously, often in new ways that are still poorly mapped and understood. The HLEG acknowledges that, while not necessarily illegal, disinformation can nonetheless be harmful for citizens and society at large. The risk of harm includes threats to democratic political processes, including integrity of elections, and to democratic values that shape public policies in a variety of sectors, such as health, science, finance and more. In light of these considerations, the HLEG points out that disinformation problems can be handled most effectively, and in manner that is fully compliant with freedom of expression, free press and pluralism, only if all major stakeholders collaborate. In addition, continuous research, increased transparency and access to relevant data, combined with regular evaluation of responses, must be permanently ensured. This is particularly important as disinformation is a multifaceted and evolving problem that does not have one single root cause. It does not have, therefore, one single solution. The HLEG advises the Commission to disregard simplistic solutions. Any form of censorship either public or private should clearly be avoided. Fragmentation of the Internet, or any harmful consequences for its technical functioning should also be avoided. The HLEG’s recommendations aim instead to provide short-term responses to the most pressing problems, longer-term responses to increase societal resilience to disinformation, and a framework for ensuring that the effectiveness of these responses is continuously evaluated, while new evidence-based responses are developed. The multi-dimensional approach recommended by the HLEG is based on a number of interconnected and mutually reinforcing responses. These responses rest on five pillars designed to: 1. enhance transparency of online news, involving an adequate and privacy-compliant sharing of data about the systems that enable their circulation online; 2. promote media and information literacy to counter disinformation and help users navigate the digital media environment; 3. develop tools for empowering users and journalists to tackle disinformation and foster a positive engagement with fast-evolving information technologies; 4. safeguard the diversity and sustainability of the European news media ecosystem, and 5 6 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION 5. promote continued research on the impact of disinformation in Europe to evaluate the measures taken by different actors and constantly adjust the necessary responses. The HLEG calls on the European Commission to consider, in its upcoming Communication on “fake news” and online disinformation, a multi-dimensional approach based on these five pillars and consisting of concrete, inter-dependent actions. Its main features are discussed in detail in Section 4 of this Report and can be outlined as follows. For the short to medium term, the HLEG suggests, as a first step, a self-regulatory approach based on a clearly defined multi-stakeholder engagement process, framed within a binding roadmap for implementation, and focused on a set of specific actions. All relevant stakeholders, including online platforms, news media organisations (press and broadcasters), journalists, fact-checkers, independent content creators and the advertising industry, are called upon to commit to a Code of Practices. This Code should reflect stakeholders’ respective roles and responsibilities. The intent should be to promote an enabling environment for freedom of expression by fostering the transparency and intelligibility of different types of digital information channels. In particular, the HLEG has formulated 10 key principles to be enshrined in this Code, which define clear objectives for platforms. To make sure that the necessary steps will be taken, the HLEG recommends establishing a Coalition representing the relevant stakeholders for the purpose of elaborating such a Code of Practices and ensuring its implementation and continuous monitoring and review. As a second step, the Commission is invited to re-examine the matter in Spring 2019 and decide, on the basis of an intermediate and independent evaluation of the effectiveness and efficiency of these measures, whether further actions should be considered for the next European Commission term. This may cover options for additional fact-finding and/ or policy initiatives, using any relevant instrument, including competition instruments or other mechanisms to ensure continuous monitoring and evaluation of the implementation of the Code These good practices must be backed by a structured crossborder and cross-sector cooperation involving all relevant stakeholders, in order to foster transparency, algorithm accountability and public trust in media to an appreciable extent.. Given the fragmentation of the sector, public authorities should play a facilitating role. Therefore, the HLEG asks public authorities, both at the EU and national level, to support the development of a network of independent European Centres for (academic) research on disinformation. This network should be open to fact- and source-checkers, accre- dited journalists, researchers from different relevant fields and platforms, with a view to: ▶▶ continually monitoring the scale, techniques and tools, and the precise nature and (potential) impact of disinformation in society; ▶▶ assessing the veracity of factual claims underpinning news and information across areas of general interest (public affairs and politics, health, science, education, finance, etc.); ▶▶ identifying and mapping disinformation sources and mechanisms that contribute to their digital amplification; ▶▶ providing a safe space for accessing and analysing platforms’ data and for a better understanding of the functioning of algorithms; ▶▶ contributing to the development of fair, objective and reliable indicators for source transparency; ▶▶ sharing knowledge with news media and platforms to enhance public awareness about disinformation. In this context, the Commission should also consider establishing a Centre of Excellence, which would act independently and with full autonomy. Its goal should be to manage the infrastructure necessary to enable an effective networking of such national research centres and to ensure a wide dissemination of their independent research outcomes. The commitment required from public authorities involves the provision of sufficient independent funding for both the establishment and operation of such a network, and a renewed effort to support R&D activities in critical fields. Examples include advanced content verification tools for newsrooms, artificial intelligence and big data for media. However, while necessary, these actions alone would not be sufficient to “dilute” disinformation with trustworthy content made more visible and findable online. Additional measures aimed at strengthening societal resilience in the longer term need to be implemented in parallel. Therefore, the HLEG recommends a set of complementary measures. These measures are designed to support the diversity and sustainability of the news media ecosystem on the one hand. On the other hand, they are designed to develop appropriate initiatives in the field of media and information literacy to foster a critical approach and a responsible behaviour across all European citizens. The HLEG chiefly calls on EU and national authorities to step up their efforts in these two areas, this also requires the cooperation of civil society organisations, media organisations and platforms. In Section 5 of the Report, the specific actions that should be implemented to pursue these longer-term objectives are listed by reference to the role that public and EXECUTIVE SUMMARY private actors are called upon to play, taking into account their respective responsibilities. More detailed information on good practices, underlying fundamental principles and recommended actions can be found in Sections 2, 3 and 4 of this Report. 7 Table of Contents Table of Contents Table of Contents ................................................................................................................... 9 1. Problem definition and scope of the Report ............................................... 10 2. Measures already taken by various stakeholders .................................. 14 3. Key principles and general, short- and long-term objectives ........ 18 4. Responses and actions ............................................................................................... 22 a. Transparency................................................................................................................. 22 b. Media and information literacy............................................................................ 25 c. Empowerment of users and journalists........................................................... 27 d. Diversity and sustainability of the news media ecosystem.................. 29 e. Process and evaluation............................................................................................ 31 5. Conclusions: summary of actions ...................................................................... 35 ANNEX I  .........................................................................................................................39 9 10 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION 1. Problem definition and scope of the Report The move to an increasingly digital environment gives European citizens many new ways of expressing themselves and of finding and accessing diverse information and views. It also enables an increase in the volume of various kinds of disinformation in circulation. The latter potentially represents risks for our democratic processes, national security, social fabric, and can undermine trust in the information society and confidence in the digital single market. The European Union and many of its Member States face these risks from a position of strength in several ways, with generally well-educated and digitally connected populations, strong political institutions, diverse news media, a competitive economy, and a vibrant civil society. However, disinformation represents risks that we have to, collectively, confront and contain to realize the full democratic, societal, and economic potential of technological progress whilst respecting the freedom of expression and the freedom to receive and impart information. The threat is disinformation, not “fake news” Current debates about ‘fake news’ encompass a spectrum of information types. This includes relatively low-risk forms such as honest mistakes made by reporters, partisan political discourse, and the use of click bait headlines, to high-risk forms such as for instance foreign states or domestic groups that would try to undermine the political process in European Member States and the European Union, through the use of various forms of malicious fabrications, infiltration of grassroots groups, and automated amplification techniques. In this report, we focus specifically on problems associated with disinformation. We define it as false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit. The risk of harm includes threats to democratic political processes and values, which can specifically target a variety of sectors, such as health, science, education, finance and more. It is driven by the production and promotion of disinformation for economic gains or for political or ideological goals, but can be exacerbated by how different audiences and communities receive, engage, and amplify disinformation. Misinformation, defined as misleading or inaccurate information shared by people who do not recognize it as such, is not our focus. In this report, the HLEG deliberately avoid the term ‘fake news’. The HLEG do this for two reasons. Firstly the term is inadequate to capture the complex problem of disinformation, which involves content that is not actually or completely “fake” but fabricated information blended with facts, and practices that go well beyond anything resembling “news” to include some forms of automated accounts used for astroturfing, networks of fake followers, fabricated or manipulated videos, targeted advertising, organized trolling, visual memes, and much more. It can also involve a whole array of digital behaviour that is more about circulation of disinformation than about production of disinformation, spanning from posting, commenting, sharing, tweeting and re-tweeting etc.(1) Secondly, the term ‘fake news’ is not only inadequate, but also misleading, because it has been appropriated by some politicians and their supporters, who use the term to dismiss coverage that they find disagreeable, and has thus become a weapon with which powerful actors can interfere in circulation of information and attack and undermine independent news media. Research has shown2 that citizens often associate the term ‘fake news’ with partisan political debate and poor journalism broadly, rather than more pernicious and precisely defined forms of disinformation. Disinformation as defined here includes forms of speech that fall outside already illegal forms of speech, notably defamation, hate speech, incitement to violence, etc. but can nonetheless be harmful. It is a problem of actors — state or nonstate political actors, for-profit actors, citizens individually (1) Wardle, C. & Derakhshan, H. (2017) Information Disorder: Toward an Interdisciplinary Framework for Research and Policy Making. Report to the Council of Europe. https://shorensteincenter.org/information-disorder-framework-for-research-and-policymaking/ (2) Nielsen, Rasmus Kleis, and Lucas Graves. 2017. ““News You Don’t Believe”: Audience Perspectives on Fake News.” Oxford: Reuters Institute for the Study of Journalism. 1. PROBLEM DEFINITION AND SCOPE OF THE REPORT or in groups — as well as infrastructures of circulation and amplification through news media, platforms, and underlying networks, protocols and algorithms. In the near future, it will increasingly involve communication via private messaging applications, chat bots, and voice-operated systems, as well as augmented reality and virtual reality and content generated or manipulated by AI. It is a problem that must be understood in the wider context of how information is produced, how it is distributed, and how people engage with it in the public sphere. This includes changes in the business and profession of journalism, the move to digital media and rise of platforms(3), as well as the widespread crisis of confidence between much of the public and many public institutions at national or EU level.(4) Therefore, the present Report addresses all forms of false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit. It does not deal with issues arising from the creation and dissemination online of illegal content, which are subject to regulatory remedies under EU or national laws, nor with other forms of deliberate but not misleading distortions of facts such as satire and parody. Problems of disinformation in the European Union Disinformation is a multifaceted problem, does not have one single root cause, and thus does not have one single solution. Some forms of disinformation have clearly been enabled by the development of specific digital media, including platform products and services, but the problem also involves some political actors, news media, and civil society actors. Problems of disinformation are thus connected with wider political, social, civic and media issues in Europe. Many of these fundamental issues predate the move to digital media and include the following four aspects. The evolving nature (3) Platforms” is a broad term which encompasses a range of activities including social media, search engines, news aggregators, marketplaces, communications services, creative content outlets, app stores, payment systems, and platforms for collaborative projects. They share key characteristics including the use of information and communication technologies to facilitate interactions between users as (information) intermediaries, collection and use of data about these interactions, and network effects which make the use of the platforms with most users most valuable to other users. In the context of the spread of disinformation, it has been shown that a wide range of types of platforms can be used. Ultimately, these distinctions are crucial to acknowledge, as solutions will need to be tailored to the specific type of platform. While the shorthand “platform” will be used throughout this paper, we acknowledge that solutions explored will need to be tested and tailored to suit the specific type of platform or service, and reference should be made to high level principles which can be adhered to by the range of actors in this ecosystem. (4) Newman, N., Fletcher, R., Kalogeropoulos, A. Levy, D.A. and Nielsen, R.K. (2017) Reuters Institute Digital News Report 2017. Oxford: Reuters Institute for the Study of Journalism. http://www.digitalnewsreport.org/ of disinformation, dissemination patterns and techniques requires a regularly updated and evidence-based understanding of the scale, scope and impact of the problem in order to calibrate appropriate responses. First, political actors can be purveyors of disinformation: for instance, foreign governments or domestic groups could be working actively to undermine the integrity of European media systems and political processes. Furthermore, domestically, not all European politicians and public authorities share the same level of respect for media freedom and independence; some actively seek to directly or indirectly control both private sector and public service news media, and some European citizens regard political actors and public authorities with considerable scepticism. Second, not all news media maintain the same standards of professionalism and editorial independence and while news media can play an important role in combating disinformation and increasing societal resilience, some news media contribute to disinformation problems, thereby weakening European citizens’ overall trust in media. It is important therefore to strengthen professional and independent media and journalism that brings societal attention to disinformation and underpins the democratic process. Third, while civil society actors play an important watchdog role in many areas, including non-profit journalism, factchecking and verification, consumer protection, media literacy training, as well as by holding political and economic powers accountable for their actions, it is also clear that some problems of disinformation are animated by citizens individually or collectively sharing false and misleading content and that highly polarized societies with low levels of trust provide a fertile ground for the production and circulation of ideologically motivated disinformation.(5) Fourth, the role of digital media, and in particular of large US-based platform companies offering useful and popular products and services for search, social networking, messaging, and the like, is important but not yet well understood. As more and more European citizens turn to platform products and services to find and access information and engage with public affairs, platform companies are becoming increasingly important as both enablers and gatekeepers of information. The growing power of platforms to enable — and potentially to interfere with — the free circulation of information comes with growing responsibilities. The tools they provide empower European citizens by creating new ways of imparting and receiving diverse information and views from a va- (5) Weeks, B. E. (2015) Emotions, partisanship, and misperceptions: How anger and anxiety moderate the effect of partisan bias on susceptibility to political misinformation. Journal of Communication. 65 (4), 699-719. 11 12 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION riety of sources than ever before.(6) However, they have also enabled the production and circulation of disinformation, on a larger scale than previously, often in new ways that are still poorly mapped and understood.(7) It is clear that many of the tools integral to the contemporary digital ecosystem that are used for legitimate purposes— e.g. behavioural data collection, analytics, advertising exchanges, tools for cluster detection and tracking social media sentiment, and various forms of AI/machine learning—have also been harnessed by some purveyors of disinformation. The threats represented by various forms of disinformation are thus embedded in a much larger, complex, and evolving ecosystem that is often opaque—sometimes by design— underlining the importance of joint efforts to make digital media, including platform companies, more transparent and intelligible so that European citizens can make informed choices about what information they access and how they engage. Special attention should be paid to the threat represented by disinformation aimed at undermining the integrity of elections (local, national or EU elections). During elections, citizens require accurate information to carry out successfully their democratic right to vote. This includes accurate information about their eligibility to vote (in terms of their personal voter registration information), the date and location of the vote, the security of the voting systems, and the methods through which they can vote. In addition, citizens require accurate information about the candidates that are running and the positions they hold on different policies, therefore, a fair playing field for presenting and communicating these policy positions should be provided for. Social media have made communication around elections much easier and more efficient, both for public authorities and citizens. Unfortunately, elections also can become vulnerable to digital disinformation campaigns.8 These vulnerabilities can include disinformation about voting time and location, rumours about rigging and hacking of voting machines, purchases and non-transparent dissemination of political advertisement. (6) Dutton, William H. and Reisdorf, Bianca Christin and Dubois, Elizabeth and Blank, Grant, Social Shaping of the Politics of Internet Search and Networking: Moving Beyond Filter Bubbles, Echo Chambers, and Fake News (March 31, 2017). Quello Center Working Paper No. 2944191. Available at SSRN: https://ssrn.com/abstract=2944191. (7) Fletcher, Richard, Alessio Cornia, Lucas Graves, and Rasmus Kleis Nielsen. 2018. “Measuring the Reach of ‘Fake News’  and Online Disinformation in Europe.” Oxford: Reuters Institute for the Study of Journalism. https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2018-02/Measuring%20 the%20reach%20of%20fake%20news%20and%20online%20distribution%20in%20Europe%20CORRECT%20FLAG.pdf (8) See https://www.intelligence.senate.gov/hearings/open-hearing-socialmedia-influence-2016-us-elections. Researchers are still examining the potential impact of these types of digital advertisements in terms of voter behaviour and general suspicion cast over the democratic process. This examination of the methods and potential impact is both important and urgent, since a breach of constitutional integrity represents a risk for democracy. Equally important is the threat of more insidious and low profile disinformation strategies which are not linked to any political event. By creating repeated distortions impacting citizens’ perceptions of events, these can give rise to deepseated misinformed beliefs and cause significant harm. The fact that these streams of disinformation are potentially less recognisable and harder to track compounds their potential damage. The questions raised by integrity of elections are therefore similar to those about integrity of information: transparency, reliability, findability and trust. All responses should avoid interference with freedom of expression and freedom to receive and impart information. Responding to disinformation As Europe faces the risks represented by various forms of disinformation, it is first of all important to ensure an updated and evidence-based understanding of the scale, scope, and precise nature of the problems at hand and to design possible responses which take into account that, because disinformation problems are deeply intertwined with the digital ecosystem, the technologies, the tactics and techniques will continue to evolve. To ensure appropriate, efficient, and future-proof responses, we will need to continually examine the problem and evaluate our responses. The aim of this report is to identify key principles and objectives for future action, and translate them into specific, measurable, achievable, realistic and time-framed responses. Drawing from the contributions provided by the different experts who took active part in the reflections of the HLEG, the following sections will first review the measures already taken by various stakeholders and assess their advantages and possible limitations. Secondly, the present Report will recall the fundamental principles in the light of the relevant case law and, taking into account stakeholders’ practices, it will set out a number of general and specific objectives for future actions. Thirdly, it will identify possible responses stemming from such principles and objectives. As disinformation is a multifaceted problem, which cannot be addressed by a single measure, the proposed responses should be seen as a set of inter-dependent actions forming part of an overarching, multi-dimensional approach. 1. PROBLEM DEFINITION AND SCOPE OF THE REPORT As we continue to move towards a more digital environment, it is clear that the same technologies and platforms can both enable entirely legitimate information, even if sometimes contentious, and perhaps poor quality, and activities that allow various forms of potentially harmful disinformation. This report aims to identify ways of increasing European societies’ resilience to threats of disinformation while maintaining an open environment that enables circulation of ideas and information and identifies targeted responses to specific problems of disinformation without undermining the ways in which digital media are empowering European citizens, societies, and economies. 13 14 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION 2. Measures already taken by various stakeholders As stated earlier, digital disinformation is a multifaceted problem, which does not have one single root cause and thus does not have one single solution. With this in mind, this Section lists some of the already existing responses undertaken by platforms, news publishers, broadcasters, fact-checkers and civil society. The references to concrete examples are by no means meant to be exhaustive. Purely governmental responses are not included. The good practices examined here are those that avoid chilling effects on freedom of expression and combat disinformation while ensuring other concerns like data privacy. By contrast, bad practices tend to present risks to freedom of expression and other fundamental rights. They include censorship and online surveillance and other misguided responses that can backfire substantially, and that can be used by purveyors of disinformation in an “us vs. them” narrative that can de-legitimize responses against disinformation and be counter-productive in the short and long run. Attention should also be paid to lack of transparency and to the privatization of censorship by delegation to specific bodies/entities or private companies. Good practices tend to fall into three major categories, transparency, trust-enhancement, and media and information literacy. We list a range of initiatives here to illustrate what different stakeholders are already doing, but it is important to note that most of these initiatives have not yet been independently evaluated, and that their efficiency is therefore not yet established, just as any potential unintended side effects are unclear. In most cases, independent evaluation would require the stakeholders behind a given initiative to share data with outside researchers. (i) Transparency and accountability-enhancing practices The most frequently deployed types of intervention tend to challenge disinformation by producing initiatives that help create resilience among citizens and empower the various actors impacted. These initiatives need to be context-specific and context-sensitive and continuously evaluated as responses that work in one context may not work in others. Examples include initiatives to influence “findability”, privileging credible content in ranking algorithms, initiatives to identify and document disinformation sources, policies aimed at ensuring an enabling environment for news media and professional journalism, as well as investments in media and information literacy aimed at fostering the intelligibility of digital media. Initiatives more focused on containing disinformation include steps to limit its spread through online interactions (sharing, commenting, liking, etc.) and organized network structures (re-shares, re-posting, sometimes driven by automated or inauthentic accounts). Other initiatives are aimed at increasing transparency around disinformation delivered through advertising and sponsored content and include enforcing advertising policies. Online platforms are making efforts to provide responses to the distribution of disinformation. Key efforts include, first, steps to identify and remove illegitimate accounts; second, steps to integrate signals for credibility and trustworthiness in ranking algorithms and include recommendations of alternative content to increase the “findability” of credible content; third, attempts to de-monetize for-profit fabrication of false information and, fourth, collaboration with independent source and fact-checking organisations. Different platforms have taken different initiatives and not all platforms have invested the same efforts and resources in containing disinformation. Importantly, many of these initiatives are only taken in a small number of countries, leaving millions of users elsewhere more exposed to disinformation. Furthermore, because of the dearth of publicly available data, it is often hard for independent third parties (fact-checkers, news media, academics, and others) to evaluate the efficiency of these responses. (9) (9) Twitter is updating its policies for electioneering advertisers to include stricter requirements on who can serve these ads and limit targeting options, require electioneering advertisers to self-identify as such, and introduces stronger penalties for advertisers who violate policies https:// blog.twitter.com/official/en_us/topics/product/2017/New-TransparencyFor-Ads-on-Twitter.html. 2. MEASURES ALREADY TAKEN BY VARIOUS STAKEHOLDERS Building on fact-checking outcomes, online platforms have started to tackle disinformation by disrupting the business model for its production and amplification. Advertising networks (operated by the platforms themselves or by third parties) play an important role within this strategy (“followthe-money”), which in essence pursues three aims: ▶▶ Advertising networks not placing ads on websites identified as purveyors of disinformation. This directly reduces the income to disinformation providers. ▶▶ Advertising providers not accepting ads from disinformation sources and clearly marking political ads as sponsored content to create transparency. ▶▶ Advertising networks not disbursing revenues to sites and partners until they have been able to confirm that they operate within relevant terms and conditions. Such steps make it harder for disinformation providers to profit from distribution. In this respect, platforms suggest involving the online advertising industry in order to develop a more comprehensive strategy. (10) Google introduced fact-check labels to Google News to allow publishers to highlight fact-checked content and help users find and consult more easily articles that provide a critical outlook on claims made by others. This feature helps support the work of the fact-checking community. This fact-checking feature first appeared in the UK and the US in October 2016 and has since been rolled out globally. Google has expanded the fact-checking labels to Google Search results, to allow publishers to provide greater awareness to users of the tools available to them to consider the accuracy of a story. These labels in Search make it easier for publishers to highlight their fact-checking work that shows users the origin of a claim and clearly display their verdict on the veracity of the claim. This work has been done in collaboration with the fact-check community, and started with sharethefacts.org a collaboration between the Duke University Report’s Lab and Jigsaw, a team within Alphabet, Google’s parent company. Share the Facts enables fact-checkers to more easily share the claims they looked at and their fact-check findings, and also makes it easier for others to highlight their fact checks, for example in Search results. Facebook has looked at the impact of “Disputed” flags and concluded that they can be counter-productive, which led to its decision to test a new approach based on the automatic display of alternative recommended content (“Related Articles”).https://newsroom.fb.com/news/2017/12/newsfeed-fyi-updates-in-our-fight-against-misinformation/ (10) In view of the specific threats posed by disinformation targeted at electoral processes, some platforms, and in particular Google, Twitter and Facebook, are conducting self-assessments and analysis of threats to their own security. They also address data automation issues linked with denial of service or tampering with their services. Facebook has developed an election integrity programme to act against hacking and malware, to examine the role of ads and foreign interference and to understand fake accounts. Together with other platforms, they are strengthening their advertising policies to provide transparency, and restrict ad content with stricter guidelines. They are also expanding partnerships with election commissions in different countries. Facebook has developed the ThreatExchange program, a social platform for sharing cybersecurity threats and information with tech companies on attempts at interference. Google Project Shield provides News, Human Rights, and Elections Monitoring sites with protection from DDoS (Distributed Denial of Service). Twitter uses API-based tools to prevent multiple accounts Likes, Retweets and Follows. Mark Zuckerberg, Public statement on election integrity, available at: https://www.facebook.com/ zuck/posts/10104052907253171 ; see also https://newsroom.fb.com/ For their part, both news publishers and broadcasters are in the process of strengthening their fact-checking capabilities - set aside the fact that both press publishers and broadcasters are subject to right of reply and are accountable by law for any inaccurate news reporting. As regards the print press, this type of activities are either integrated within the newsrooms’ investigative departments or may involve media assessment NGOs, notably those utilizing data driven journalism to identify, track, uproot disinformation and its sources. (11) Independent source and fact-checkers and civil society organisations. European source and fact-checking culture is still being developed with some promising efforts in bigger Member States(12). Reach and speed are key issues for fact-checkers. Current operations are probably not reaching a large enough audience fast enough to counter viral falsehoods. The question is whether and how source and factchecked information could benefit from wider visibility. More work can and should be done by fact- and source checkers in a collaborative manner within EU members states (like with CrossCheck) and across the EU, in line with sporadic examples like RefugeeCheck. news/2018/01/effect-social-media-democracy/ See https://developers.facebook.com/products/threat-exchange See projectshield.withgoogle.com/public See https://blog.twitter.com/developer/en_us/topics/tips/2018/automation-and-the-use-of-multiple-accounts.html (11) Fact-checking has yet to become a mainstay of broadcasting formats across the continent, even as some experiments such as Channel 4 Fact Check, BBC Reality Check and the collaboration between RAI 2 and Pagella Politica have shown potential over the past years. More recently, in 2017, ahead of the German election, ARD-aktuell built up a team of journalists and researchers to focus on the issue of disinformation. They created the ‘Faktenfinder’ team aimed at debunking disinformation. It also counters targeted false information with its own research results and uses concrete examples to show how to identify incorrect claims or dubious messages. The results are used on ARD broadcasts. Faktenfinder has a mission to improve media literacy in Germany. The second example comes from the RTL Group, which has created a ‘verification team’ that includes operators from different countries plus a number of partner organizations. This team is specially trained to verify/falsify video material. They are connected via a joint e-mail address and can be reached 24/7. They also regularly test new tools that help identify fake videos and/or fake pictures. In addition to that, they offer basic verification trainings for members of all editorial teams within the Group. In Spain, Maldito Bulo, an independent project inside Maldita.es has sought to collaborate with national TV laSexta and radio Onda Cero in order to be able to have a bigger impact on the public. Maldito Bulo, carries out debunking on social media through easy readable images. It has built a database which is constantly updated and can be consulted by users, and it has also developed a plug-in that, when installed, alerts about webpages’ level of liability in relation to the number of stories debunked by MB from that webpage. (12) Graves, Lucas and Federica Cherubini. “The Rise of Fact-Checking Sites in Europe”. Oxford: Reuters Institute for the Study of Journalism. https:// reutersinstitute.politics.ox.ac.uk/our-research/rise-fact-checking-sites-europe 15 16 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION (ii) Trust-enhancing practices and algorithm changes A second category includes various types of measures that leverage on artificial intelligence and machine learning to tackle specific facets of the disinformation phenomenon and filtering systems enabling the exposure of fact-checked information. Online platforms have been experimenting with different ideas, including partnering with publishers and independent fact-checkers and developing trust measures.(13) Print press organisations and broadcasters are in the process of intensifying their efforts to enforce certain trustenhancing practices. This includes media curation pointing to good public sources in the digital environment. Journalism guidelines are issued by individual news media, international organisations such as the International Federation of Journalists, as well as national bodies. Guidelines include deontological codes, ethics and standards to guarantee quality in the methods in which news is produced. Broadcasters have taken slightly different measures than press, because of the nature of their business and the different legal frameworks in which they operate. Most print press and broadcasting organisations, however, have codes of conduct, and are obliged to grant a ‘right of reply’ to third parties. Moreover, in most countries, broadcasters are obliged to be transparent on media ownership, ensure impartiality of news, protect minors, and limit advertising and product placement. (13) For example, Facebook recently announced that it is developing new trust scores based on surveying users, while Google, Twitter, Facebook and Bing have all been working with the Trust Project . To some extent, they have also shown interest in other initiatives such as the Credibility Coalition and the Journalism Trust Initiative of Reporters Without Borders. In those platforms that use algorithms to organise content, changes in algorithms have a major impact on the online display of news. Examples include Facebook and Google, while other platforms such as Twitter have different selection logics. Google recently took steps to strengthen their understanding of content quality with a view to improving and refining their algorithms. As with all ranking changes, Google works with external Search Quality Raters to run experiments in order to gather data on the quality of the results, so they can identify areas where improvement is needed. These human raters assess how well a website gives people who click on it what they are looking for. These ratings help inform the systems that govern ranking and results, rather than having a direct effect on the particular site reviewed. They help Google benchmark the quality of the results and make sure these meet a high bar for users of Google Search all around the world. To ensure a consistent approach, Google publishes Search Quality Rater Guidelines to give Evaluators guidance and examples for appropriate ratings. In order to identify and address misinformation, Google has updated their Search Quality Rater Guidelines to explicitly address misleading content. They will use ratings against these criteria to evaluate whether changes to Search algorithms will improve the quality of results, and to make additional improvements over time.See https://newsroom.fb.com/news/2018/01/trusted-sources/ Press organisations are also looking into other measures, such as credibility indexes, to help audiences understand how news is produced and whether these processes adhere to the ethical and professional standards mentioned above. (14) As regards fact-checking organisations, quality standards exist already, namely the International Fact-Checking Network (IFCN) Code of Principles, which is signed by most major fact-checkers in the world and entails an extensive accreditation process. Still, fact-checkers must continuously improve on their transparency - and organizations that are not signatories of this code should strive to become verified. As regards more specifically the role played by algorithms in displaying and ranking news and information online, factcheckers and news media organisations add that they should as trustworthy organizations be allowed to better understand how such algorithms impact the distribution of their services and products(15) and that, in any case, major changes to algorithms should be announced in advance. Moreover, factchecking organisations claim that they cannot evaluate the impact of their work (e.g. reduced reach of the debunked piece of content via downgrading in the newsfeed, or effectiveness of the automatic display of related articles in dissuading people from sharing the link) without greater transparency. The authorisation for an appropriate use of APIs from online platforms, eventually through academic, scientific or independent body oversight, would facilitate their endeavours. Platforms and others warn that an excessive level of transparency of algorithms might create opportunities to game these systems. While platforms already provide blog posts with some updates and insights into products, policy, and algorithmic changes, it is felt that access to data in the public interest and for research purposes should be improved in a manner that is compliant with privacy requirements and respects commercial interests. (iii) Media and information literacy A third category comprises media and information literacy practices, as education of the public at large is another way (14) The InVID innovation (2016-2018) action funded by the EU under the Horizon 2020 program develops a knowledge verification platform to detect emerging stories and assess the reliability of newsworthy video files and content spread via social media. In July 2017, InVID released a verification plugin conceived as a toolbox to help journalists verify images and videos as well as debunk disinformation. This plugin has become in a few months a state-of-art tool recommended worldwide by the lead users in information verification in journalism and beyond. Hundreds of active users from the main media newsrooms, NGOs dealing with human rights, and media educators have adopted this toolbox in their verification processes. http://www.invid-project.eu/ (15) See for instance https://www.politico.com/story/2017/09/07/facebook-fake-news-social-media-242407 and https://www.politico.com/ story/2018/01/07/facebook-fake-news-326996 2. MEASURES ALREADY TAKEN BY VARIOUS STAKEHOLDERS of building resilience among citizens, end-users and voters. The purpose is to bolster prevention and to reduce the appeal of disinformation and conspiracy theories. It builds on media and information literacy’s know-how to foster critical thinking on propaganda and advertising, bolstered by Safer Internet practices for responsible social media use. Although online platforms are not the main initiators of media literacy programmes, they have started to play some role in this area. It should however be for independent (educational) institutions to provide the content of any media and information literacy programmes. Certain press media, broadcasters and Public Service Media (PSM) from their public role, carry out specific media literacy projects, mostly in cooperation with schools and other educational institutions, notably targeting the young public. Civil society organisations are very active in developing media literacy actions and programmes including via experimental collaborations with other stakeholders, to help the younger generations to become conscious consumers of news within the new digital ecosystems. Moreover, many interesting initiatives are already undertaken to help older generations improve their media literacy skills. 547 of such projects are featured in a report by the Audiovisual Observatory published in March 2017.(16) Many organisations active in this field point out, however, that for media and information literacy to be effective, it should be implemented on a massive scale in school and university curricula and in teacher training curricula, with clear methods of evaluation and cross-country comparison and with reflection in educational rankings gauges. (16) See https://www.obs.coe.int/en/web/observatoire/reports 17 18 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION 3. Key principles and general, shortand long-term objectives Freedom of expression The European Union, its Member States, and all relevant stakeholders should face the risks represented by various forms of disinformation within the framework provided by the European Union Charter of Fundamental Rights of the European Union and the European Human Rights Convention, as well as more broadly joint frameworks, directives, and regulations dealing with hybrid threats, the digital single market, and European and Member States media regulation. Freedom of expression is enshrined in Article 11 of the Charter of Fundamental Rights of the European Union as an indispensable enabler of sound decision-making in free and democratic societies.(17) Freedom of expression extends to print, broadcast and online media and includes the right to hold opinions and to receive and impart information and ideas “without interference by public authorities and regardless of frontiers,”(18) as well as the integral, corollary values of media freedom and media pluralism.(19) Freedom of expression is part of the national legal and democratic order in all EU Member States. It is likewise safeguarded by European and international human rights treaties to which all of the Member States are parties – notably, the European Convention for the Protection of Human Rights and Fundamental Freedoms (Convention) of the Council of Europe(20) and the International Covenant on Civil and Political Rights (Covenant) of the United Nations.21 In construing the right to freedom of expression under Article 10 of the Convention, the European Court of Human Rights (the Strasbourg Court) has emphasised the “essential function the press fulfils in a democratic society” notably its “public watchdog” role. At the same time, the Strasbourg Court has noted that “the safeguard afforded by Article 10 to journalists in relation to reporting on issues of general interest is subject to the proviso that they are acting in good faith and on an accurate factual basis and provide ‘reliable and precise’ information in accordance with the ethics of journalism.“(22) Under the Charter, any limitations on freedom of expression must be prescribed by law, proportional, and necessary either to meet general interest objectives recognised by the Union or to protect the rights and freedoms of others.(23) The Convention(24) and Covenant(25) establish broadly parallel requirements. The Union has adopted or proposed a number of legislative measures on illegal content aimed at meeting general interest objectives and Member States enforce national measures covering content that is defamatory or injurious to reputation. Illegal content does not need to be considered for the purposes of this report but is mentioned here pro memori. (22) Stoll v. Switzerland [GC], no. 69698/01, ECHR 2007-XIV, §, and cases cited therein. (23) Charter, Article 52(1). (17) Article 6(1) of the Treaty on European Union (TEU) states that “[t]he Union recognizes the rights, freedoms and principles set out in the Charter …, which shall have the same legal value as the Treaties.” (18) Article 11(1), Charter. (19) Article 11 (2), Charter. (20) Article 10, Convention. Article 52 (3) of the Charter states that “[i[n so far as this Charter contains rights which correspond to rights guaranteed by the [Convention], the means and scope of those rights shall be the same as those laid down by said Convention. This provision shall not prevent Union law providing more extensive protection.” TEU Article 6(3) states that, “fundamental rights, as guaranteed by the [Convention] and as they result from the constitutional traditions common to the Member States, shall constitute general principles of the Union’s law.” (21) Article 19, Covenant. (24) Article 10(2) of the Convention provides that the exercise of freedom of expression, “since it carries with it duties and responsibilities, may be subject to such formalities, conditions, restrictions or penalties as are prescribed by law and are necessary in a democratic society, in the interests of national security, territorial integrity or public safety, for the prevention of disorder or crime, for the protection of health or morals, for the protection of the reputation or rights of others, for preventing the disclosure of information received in confidence, or for maintaining the authority and impartiality of the judiciary.” (25) Article 19(3) of the Covenant provides that the exercise of the right of freedom of expression “carries with it special duties and responsibilities. It may therefore be subject to certain restrictions, but these shall only be such as are provided for by law and are necessary (a) For the respect of the rights or reputations of others; (b) For the protection of national security or of public order (ordre public), or of public health or morals.” 3. KEY PRINCIPLES AND GENERAL, SHORT- AND LONG-TERM OBJECTIVES International dimension of disinformation and evolving international law standards European and international jurisprudence and standards recognise that positive action by States may be required to secure the effective exercise of freedom of expression. In its jurisprudence, the Strasbourg Court has underlined the positive obligation of States under Article 10 of the Convention to enable “an environment favourable to all for participation in public debate,” in particular as regards the protection of journalists and media workers.(26) Texts adopted by the standards-setting bodies of the Council of Europe also cite the positive obligation of its member states to foster a favourable environment for freedom of expression, including the adoption of legislative and policy frameworks that inter alia encourage and support media pluralism and diversity of content, ensure the transparency of media ownership, organisation and funding, and provide media and digital literacy to citizens and users.(27) While not legally binding, these standards indicate the direction in which these pertinent bodies of the Council of Europe believe law and policy should develop. The 2017 Joint Declaration on “Fake News,” Disinformation and Propaganda (Joint Declaration),(28) adopted by the (Special Rapporteurs,(29) provides the most focused, recent (26) Dink v. Turkey, No. 2668/07 & others, para. 138 (14 December 2010) («Elle estime aussi que les obligations positives en la matière impliquent, entre autres, que les Etats sont tenus de créer, tout en établissant un système efficace de protection des auteurs ou journalistes, un environnement favorable à la participation aux débats publics de toutes les personnes concernées ….» Fuentes Bob v. Spain, No. 29293, para. 38 (29 February 2000); Ozgür Güdem v. Turkey, No. 23144/93, paras. 42-43 (ECHR 2000-III0. (27) See, e.g., Recommendation CM/Rec(2016) of the Committee of Ministers to member states on Internet Freedom (13 April 2015) (calling on member states to create an enabling environment for Internet freedom, including inter alia the provision of media and digital literacy programmes). https://search.coe.int/cm/Pages/result_details.aspx?ObjectId=09000016 806415fa. See also, e.g., Draft Recommendation of the Committee of Minsters to member states on media pluralism and the transparency of media ownership, MSI-MED (2016) 09rev6 (Sixth revised draft as of 7 December 2017) (reaffirming and supplementing existing COE standards on media pluralism and media transparency) http://rm.coe.int/draft-recommendation-on-media-pluralism-and-transparency-of-media-owne/168077351e. See also Media and Internet Division, Directorate General of Human Rights and Rule of Law, Council of Europe, Recommendations and Declarations of the Committee of Ministers of the Council of Europe in the field of media and information society at https://rm.coe.int/1680645b44 (28) Joint Declaration on “Fake News,” Disinformation and Propaganda, The United Nations (UN) Special Rapporteur on Freedom of Opinion and Expression, the Organization for Security and Co-operation in Europe (OSCE) Representative on Freedom of the Media, the Organization of American States (OAS) Special Rapporteur on Freedom of Expression and the African Commission on Human and Peoples’ Rights (ACHPR) Special Rapporteur on Freedom of Expression and Access to Information (3 March 2017), http://www.osce.org/fom/302796?download=true (29) The Special Rapporteurs act pursuant to specialised mandates from their respective intergovernmental bodies – the UN, the OSCE, the OAS and the treatment of the application of international human rights standards to the phenomenon of disinformation. The Joint Declaration underlines inter alia the potential of disinformation and propaganda to mislead and interfere with “the public’s right to know as well as the right of individuals to seek and receive, as well as to impart, information and ideas of all kinds.”(30) It articulates general principles and standards and defines roles and responsibilities for States, digital intermediaries, and media outlets. In particular, the Joint Declaration highlights the “positive obligation” of States to create an “enabling environment for freedom of expression,” and identifies broad strands of public policy to this end.(31) Framework for facing disinformation Both the Charter and the Convention affirm Europe’s particular constitutional commitment to freedom of expression and the right to receive and impart information and ideas without interference by public authority and regardless of frontiers. This represents a specifically European constitutional dimension to debates around disinformation and provides the framework for setting out general and specific objectives that should inspire possible lines of actions to respond to the challenges raised by the online spread of disinformation that is harmful for society and citizens, but not illegal in nature. In terms of general objectives, it follows from the above that the aim of responses should be: ▶▶ to increase the long-term resilience of EU citizens, communities, news organisations, Member States and the EU as a whole to empower Europeans and help them proactively recognize various forms of disinformation, and ▶▶ to ensure that responses to disinformation are always upto-date, which requires to constantly monitor the evolving nature of the problems at hand, continuously innovate in designing adequate responses, and evaluate their efficiency. Based on these general objectives, most of the responses will be of a non-regulatory character and involve a wide range of different stakeholders as government or EU reguACHPR – to promote international cooperation and articulate international standards relating to freedom of expression, media freedom and media pluralism. While the Joint Declarations are not legally binding texts, they constitute persuasive interpretations of human rights standards on the themes they address. (30) Joint Declaration, fourth preambular paragraph. (31) See Joint Declaration, at Point No. 3. These include inter alia (i) a clear regulatory framework protecting broadcasters against political or commercial interference; (ii) an independent and resourced public service media; (III) measures to support media diversity, including (as warranted) subsidies or other support for the production of diverse, quality content; (iv) rules addressing media concentration and transparency of media ownership; and (v) programmes to support media and digital literacy. 19 20 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION lation of disinformation can be a blunt and risky instrument. Regulatory responses need to ensure due process, as well as accountability and proportionality. The HLEG believes the best responses are likely to be those driven by multistakeholder collaborations, minimize legal regulatory interventions, and avoid the politically dictated privatization of the policing and censorship of what is and is not acceptable forms of expression.(32) They should be based on clearly defined principles established through a transparent multistakeholder engagement process and framed within a binding roadmap for implementation including monitoring and reporting requirements in order to make them effective in practise. These responses, by all stakeholders, should not lead to harmful consequences for the technical functioning of the Internet; among others, they should avoid its fragmentation, and ensure that its security, stability and resiliency is intact. Initiatives aimed at increasing European societies’ resilience through media and information literacy, digital citizenship, stronger independent news media, and digital debate free from interference from public authorities and powerful private actors can be broad and will be broadly beneficial because they will strengthen our societies overall and increase our ability to resist various forms of disinformation. Initiatives aimed at countering specific problems of disinformation, on the other hand, while clearly potentially valuable, need to be very precisely targeted and formulated to ensure that they do not by accident or design enable public or private authorities to restrict free speech. These general objectives should be pursued based on the recognition that information can rarely be neatly categorized as simply true or false, the conviction that no authority has a monopoly on knowledge, and the belief that free societies benefit from -- and are defined by the acceptance of -- free confrontation of diverse and sometimes uncomfortable ideas and information.(33) Moreover, taking into account the best practices already applied by various stakeholders, such general objectives can be further broken down in the following specific objectives: ▶▶ Public authorities should provide an enabling environment for substantial media pluralism through a combination of de facto and de jure protection of basic rights to free expression and diverse information, including appropriate forms of support for private sector media, and support for demonstrably independent public service media who (32) See e.g. Belli, Luca, David Erdos, Maryant Fernández Pérez, Pedro Augusto P. Francisco, Krzysztof Garstka, Judith Herzog, Krisztina Huszti-Orban, et al. 2017. Platform Regulations: How Platforms Are Regulated and How They Regulate Us. FGV Direito Rio. http://bibliotecadigital.fgv.br/dspace/ handle/10438/19402. (33) See Handyside v United Kingdom (5493/72). Article 10 is applicable not only to “information” or “ideas” that are favourably received or regarded as inoffensive or as a matter of indifference, but also to those that offend, shock or disturb the State or any sector of the population.” can help produce quality information, counter disinformation, and increase media and information literacy (public authorities). Together such responses can help enhance media pluralism and diversification of the digital landscape/environment. ▶▶ Multi-stakeholder collaborations should be developed in order to independently identify, monitor, document, and alert citizens to hostile “information operations” be it from foreign states or domestic groups (especially in advance of elections). ▶▶ The production of false and harmful information for profit should be demonetized on the basis of clear, transparent, politically unbiased, and equally applied criteria. ▶▶ Platforms should allow access to data to do independent inquiries, audits and research into activities reliant on proprietary media and data infrastructures with a view to ensuring transparency and authenticity of information sources and better informed European citizens. ▶▶ Independent news media, fact-checkers and verification specialists should continue to invest in and work with platforms to build technology to more effectively monitor social streams, undertake source-checking, establish content provenance, and forensically analyse images and videos at scale and speed, to counter disinformation (including when published by news media) and to document and publicize who produces and promotes it to ensure greater transparency. ▶▶ Collaboration between news media, technology companies, civil society organizations, and researchers around the transparent development and deployment of standards of equally-applied machine-readable signals that help recognition of credible and trustworthy content, bearing in mind that technological solutions alone cannot solve social and political problems and that individuals and organizations ultimately are responsible for how technological solutions are developed and how they work. ▶▶ Educational institutions, news media, and public authorities should invest in independently run digital media and information literacy (MIL) efforts to increase awareness and understanding of media and information, digital technology, and data analytics. These efforts should be differentiated for different demographics, including not only children and young people, but also the adult population, and working to counter inequalities in how well equipped different European citizens are to make full use of digital media. Online platforms should facilitate the deployment of such efforts by developing the appropriate tools, op- 3. KEY PRINCIPLES AND GENERAL, SHORT- AND LONG-TERM OBJECTIVES timised for the specific characteristics of their services. Each stakeholder should thereby stay within its mission. ▶▶ Investment in European centres for interdisciplinary, practically useful, timely, and accessible independent evidence-based research on problems of disinformation and for rigorously evaluating the efficiency, effectiveness and appropriateness of responses deployed. The concrete actions that could be envisaged in the pursuit of these objectives are described in the next Section. 21 22 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION 4. Responses and actions In order to translate the framework for policy action set out in Section 3 above into possible responses and operational recommendations, the HLEG has deemed appropriate to focus on five intervention areas by distinguishing between actions designed to (a) enhance transparency of the online digital ecosystem, (b) promote and sharpen the use of media and information literacy approaches to counter disinformation and help users navigate our media environment, (c) develop tools for empowering users and journalists and foster a positive engagement with fast-evolving information technologies, (d) safeguard the diversity and sustainability of the European news media ecosystem, and (e) calibrate the effectiveness of the responses through continuous research on the impact of disinformation in Europe and an engagement process that includes predefined and time-framed steps combined with monitoring and reporting requirements. In line with the spirit of cooperation and engagement that has guided the dialogue within the HLEG, the following recommendations are addressed to both public and private stakeholders. a. Transparency Transparency is a key element in the response to digital disinformation. It is a cross-cutting issue that concerns the whole digital media value chain and aims at making news production and distribution less opaque with a view to supporting users’ ability to better discern between journalistic quality content and various kinds of disinformation. Furthermore, transparency may strengthen the impact of media literacy actions by providing users, both professionals and citizens, with relevant information, i.a. on patterns of circulation, necessary to critically assess the material facts quoted in news articles by journalists or in posts and blogs by citizens. Transparency can help provide more information about factual claims so that users can better evaluate the veracity of the news they access online, understand the process behind their online dissemination and popularity, the motivations and funding sources of those who make information publicly available. In this respect, platforms can immediately work with advertisers to enhance transparency by providing information on political advertising, clearly labelling content that has been paid for and is being promoted during a campaign, and making contextual information, including the originators and online amplifiers of false news, available to users and for research and transparency purposes. The journalism and academic community should continuously work for a more transparent ecosystem. In this respect, access to platforms’ data is key to a better understanding the dissemination patterns of digital disinformation. This should be an area of dedicated research work to (a) better understand scale and scope of disinformation problems and (b) evaluate efficiency of responses. This work will require an improved access to data to a wide range of legitimate third parties through appropriate APIs, ideally independently governed, while complying with privacy and data protection requirements. Point 4.a (iii) below will further discuss these aspects. Overall, the European Commission, together with Member States, should support cooperation between media organisations, platforms, academic researchers, fact-and source checkers, advertising industry and civil society organisations to ensure the necessary level of public scrutiny and balance in the definition of transparency standards. Such cooperation should in particular focus on transparency of (i) source funding, (ii) online news sources and journalistic processes and (iii) fact-checking, source-checking and verification practices. (i) Actions to increase transparency of funding sources Ownership: All digital media should provide the necessary information to help the reader to identify who is behind a certain type of information. Platforms should display this information. Sponsored content has to be clearly identifiable. Transparency in terms of political advertising and sponsorships should be encouraged as a first step. Furthermore, it is important that advertising is not inadvertently funding disinformation thereby taking away budgets for others. The “follow-the-money” approach should be strengthened with the 4. RESPONSES AND ACTIONS cooperation of the advertising industry.(34) In this connection, the «follow-the-money» approach could be explored as a possible mechanism to help ensure that ad placements are matched against on trustworthy information websites and away from sites that contribute to the creation, circulation or amplification of disinformation. Information on payments to human influencers and use of robots to promote a certain message should be made available in order for users to understand whether the apparent popularity of a given piece of online information or the apparent popularity of an influencer is the result of artificial amplification or is supported by targeted investments. While this is an area where it is difficult to obtain clear information, it is important to stress that the research work to be developed by the proposed European Centres for research on Disinformation mentioned at point 4 a (iii) below should effectively contribute to a more transparent digital ecosystem and therefore facilitate the identification of such amplification techniques. This would also contribute to effectively empower users and strengthen therefore the actions mentioned at point 4.c below. (ii) Actions to increase transparency of online news sources and journalistic processes Source transparency Indicators should be developed based on professional codes of conduct and users’ expectations for what constitutes credible and trustworthy content, adjusted to cater for online journalistic practices. Part of the solution to the spread of disinformation is to ensure that quality and trusted news is substantially available and easily findable in the digital environment. Some professional news organisations have acknowledged their role as a place where readers and viewers expect to find accurate information. They should therefore when possible continue to invest in their own journalists as well as interdisciplinary fact-checking groups and verification teams that are specialised in identifying disinformation. Moreover, they should keep up efforts to train journalists in a way that they can verify information in this changing (technological) landscape. Media organisations should continue to implement measures to reduce the risk of mistakes to the minimum when reporting news and should create transparent systems allowing the investigation of claims relating to the appearance ( ) In March Twitter will launch a transparency centre that will offer everyone visibility into who is advertising on Twitter, details behind those ads, and tools to share feedback. Specifically, the Transparency Center will show: all ads that are currently running on Twitter, including PromotedOnly ads; how long ads have been running, ad creative associated with those campaigns, targeted ads, as well as personalized information on which ads are eligible to receive based on targeting. Users can also report inappropriate ads or give negative feedback for every ad running on Twitter. 34 of false information on their services, in line with their standards and ethical guidelines. The ongoing work by Trust Project and Schema.org to create mark-ups for credible sources and for platforms can provide useful orientations to further the search for appropriate indicators. Trustworthy media organisations should therefore work together to be recognised as providers of credible information. Joint initiatives and campaigns can be used to explain the editorial standards that professional media companies need to meet and teach users how to assess the different sources of information that they can find online. Such indicators should allow users to assess where the content they access through the platforms comes from and enable them to verify the source, its ownership, and its adherence to ethical and journalistic codes. Tests and further development of such criteria should be ensured by the European Centers for Research mentioned below at point 4a (iii). Such indicators should have nothing to do with the quality of the content but rather with processes, ownership, identity and keeping a good, trusted track record (e.g. retracting incorrect stories, allowing a right to reply and rigorously enforcing a right to correction). All relevant information should be made available next to the content displayed online on whatever platform in order to empower readers to judge content trustworthiness. This will make users aware as to how to assess different sources of information. Such systems will need to be evaluated, to better understand whether flags or ‘nutritional labels’ might have unintended consequences. Platforms should integrate these source transparency indicators into their ranking algorithms in order to better signal trustworthy and identifiable sources in both search engines results and social media news feeds. These initiatives should be fully tested and evaluated before being rolled out. Dilution of disinformation with quality information - Increasing the visibility for quality media content will not only reduce the users’ exposure to low quality content but could also strengthen news media revenues. More visible quality content could also produce positive spillovers for the advertisement industry, which will see brand names associated more to quality and trusted content than to disinformation. Moreover, platforms should provide transparent and relevant information on the functioning of algorithms that select and display information without prejudice to platforms IPRs. Transparency of algorithms needs to be addressed with caution. Platforms are unique in the way they provide access to information depending on their technological design, and therefore measures to access information will always be reliant on the type of platform. 23 24 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION It is acknowledged however that, more information on the working of algorithms would enable users to better understand why they get the information that they get via platform services, and would help newsrooms to better market their services online. As a first step platforms should create contact desks where media outlets can get such information. Furthermore, it is extremely important that independent quality news media organisations that are rated as credible and trustworthy on the basis of clear criteria are informed in due time when relevant changes are made to search and news feed algorithms running the ranking of content displayed. Measures taken by platforms to empower news media and fact-checkers should be equally available across the European Union and not only available in select Member States. Transparency of algorithms is also needed for integrity of elections and this cannot be conducted without data checking. This could be done by requiring platforms to use open application programming interface, that does not reveal the algorithm but its results. This would allow third parties to build software for data checking to monitor effects. Twitter, for instance, is making its open API available to many organizations and small research groups.(35) (iii) Actions to increase transparency and efficiency of fact-checking practices Cooperation - As fact-checking activities in the EU are still relatively fragmented36, more work can and should be done by fact-checkers, verification organizations, and professional newsrooms in a collaborative manner within EU Member States and across the EU to exploit the untapped potential of cross-border and cross-sector cooperation and improve their working methods through the adoption of state-of-the-art technologies. Existing partnerships with platforms should be expanded across Europe with a clear roadmap for data sharing with academics that will allow for better understanding of disinformation strategies and their dynamics. Relevant stakeholders should engage in order to support cross-border and cross-sector cooperation between fact-checkers adhering to the IFCN Code of Principles, newsrooms, NGOs and the research world. The goal is to make sure that fact-checkers continuously improve the transparency of their working methods. Such cooperation, could build on the existing partnerships with certain online platforms in a few EU countries and be extended at the EU level, while developing innovative approaches and transparent exchanges of information, and up-taking state-of-the-art, open source technologies for fact-checking and source-verification purposes (including by (35) See https://developer.twitter.com/en/docs/tweets/search/overview (36) See 2016 Reuters Institute study on European fact-checking for more. journalists and users). The final goal should be the creation of an open market for fact-checking that avoids a “monopoly of truth” which could be potentially abused in some countries and might not carry public approval in other countries. Such cooperation could be promoted through a joint public and private effort, for instance in the form of a public/private partnership. The entity managing such a network should have a board composed of experts and operate autonomously and independently, without interference by public powers. To this end, it is suggested to support the creation of European Centres for interdisciplinary and independent evidence-based research on problems of disinformation. In order to ensure appropriate, efficient, and future-proof responses, there is a clear need to continually examine the problem and evaluate the effectiveness of the concrete measures adopted by all actors concerned. The role of such Centres should be to manage interdisciplinary research projects aimed at (a) continually monitoring the scale, technologies, tools, precise nature and (potential) impact of disinformation in society (b) assessing the veracity of factual claims underpinning news and information across areas of general interest (public affairs and politics, health, science, education, finance, etc.); (c) identifying and mapping disinformation sources and mechanisms that contribute to their digital amplification; (d) providing a safe space for accessing and analysing platforms’ data and for a better understanding of the functioning of their algorithms, (e) contributing to the development of fair, objective and reliable source transparency indicators ; (f) sharing knowledge with news media and platforms to enhance public awareness about disinformation. In organisational terms, such Centres could be run within national research organisations which, under EU law, are defined as entities (such as universities or research institutes, technology transfer agencies, innovation intermediaries, research-oriented physical or virtual collaborative entities) “whose primary goal is to independently conduct fundamental research, industrial research or experimental development or to widely disseminate the results of such activities by way of teaching, publication or knowledge transfer”(37). The multidisciplinary teams managed by the Centres could be framed as innovation hubs or “living labs” open to fact and source checkers, accredited journalists, researchers from different relevant fields and also include the participation of representatives of Press Councils and platforms. Support schemes for providing funding to research organisation operating innovation hubs or living labs are widespread across Member States and the legal conditions for such a support (37) Communication from the Commission — Framework for State aid for research and development and innovation, 2014/C 198 4. RESPONSES AND ACTION have been already clarified in the EU Framework for State aid to R&D&I(38). In order to ensure an efficient coordination of the activities of the Centres and harness synergies, the Commission should consider the possibility to set up a Centre of Excellence, acting independently and in full autonomy, and ensuring the management of a digital service infrastructure enabling the effective networking of the national centres and a wide dissemination of their research outcomes. A dedicated funding facility should be foreseen from the start to secure long-term operations. Access to data - Platforms should enable privacy-compliant access to data for the identification of online disinformation actors, for the assessment of fact-checking and debunking strategies and for the study of disinformation dynamics by academics. More data from the platforms would also cater for advertisers needs by allowing more transparency as to the placement of their ads. This would avoid misplacements and unintended funding of disinformation. (39) The HLEG recommend significant further work looking at the technical feasibility and real impact of different models. Public authorities at all EU levels should share data promptly and efficiently when it is requested by fact-checking trusted organisations. EU institutions and statistical bodies should put in place or upgrade rigorous correction policies for officials and published content so that fact checking can result in corrections more often. Furthermore, some platform companies, like Facebook and Google, have experimented with flagging links to disinformation on their products and services or highlighting factchecking work. To evaluate the effectiveness of such responses, platform companies should set out a clear timeline for sharing information on how many links are flagged, what percentage of overall content surfaced this represents, and other data that allows independent third-parties including fact-checkers and researchers to evaluate impact. To ensure these responses do not have adverse consequences, it is also important that they work on appeals processes and make clear how these appeals work. Monitoring and exchange of best practices to ensure fair and independent fact-checking, source-checking and visual verification. A good starting point would be the quality standards already existing in the field of fact-checking, namely the IFCN code of principles, which is signed by all major factcheckers in the world and requires an extensive accreditation process. Fact-checkers should clearly indicate to readers their intent and purpose. On a general note, the HLEG stresses that, in rolling out these initiatives, the EU should consider the 2019 European Parliament elections and the opportunities for launching panEuropean solutions associated to them. b. Media and information literacy In the context of the contemporary information age, media and information literacy (MIL)(40) is acquiring a strategic importance for digital citizenship as basic educational competences were for citizens of the industrial age. Media and information literacy has become an essential competence as it is the starting point for developing critical thinking and good personal practices for discourse online, and consequently also in the offline world. It aims at building a citizenship based on fundamental rights like freedom of expression, and at enabling an active and responsible participation in the online public sphere. The present information age however requires life-long learning because of the speed of change. Media literacy cannot therefore be limited to young people but needs to encompass adults as well as teachers and media professionals who often cannot keep the pace of the digital transformation induced by fast-evolving media technologies. In narrower and more immediately practical terms, media and information literacy helps to ensure that the digital information ecosystem is trustworthy: a critical readership will give an incentive to media companies to continuously improve their products and services.(41) The HLEG stresses that media literacy is an important action line as a response to disinformation because it can empower individual users as suggested above and mass empowerment of users will lead to greater social resilience against disinformation and perhaps other disorders of the information age. (38) Idem (39) Twitter offers researchers, developers, and users, access to Twitter data through its API. Twitter data is a comprehensive source of live, public conversation and enable analysis of data in real-time. Over the past months Twitter has continued to expand on its efforts to make more data available to a wider audience of researchers and SMEs, including creating a public roadmap to see what’s coming and a new developer resource centre for documentation and guidance. Twitter has also announced new API policies to limit malicious automation and potential account abuse See https://blog.twitter.com/developer/en_us/topics/tips/2018/automation-and-the-use-of-multiple-accounts.html] (40) MIL can be defined as the capacity to exert critical thinking as to the productions, representations, languages (visuals, texts, sounds), audiences and communities characteristic of mainstream and social media. See Divina Frau-Meigs and Berengere Blondeau, YALLA EMI, Un guide d’éducation aux medias et à l’information, IREX et Savoir*Devenir 2018 (forthcoming). (41) One of the key learnings of CrossCheck was that running debunks on platforms with very large audiences and writing the pieces in a way that transparently shows the process taken by journalists, audiences showed that they learned to how they could do this work themselves. 25 26 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION The strength of media and information literacy is that it is a preventive, rather than a reactive solution, engendering critical thinking skills that are crucial for the 21st century citizen living in an increasingly digital environment. These skills are all the more crucial given that biased and polarising content is often sophisticated and hard for verification systems to detect as it increasingly splices truth and fiction and circulates in hard-to-track formats.(42) tional rankings gauges.(47) This is currently missing. From a European perspective, education is an area of subsidiarity meaning that Member States have primary competence. There is a need for an incentive at international level, given the difficulties of achieving change in national education systems. The HLEG therefore proposes that media and information literacy be added to the competencies assessed in the OECD PISA rankings. The academic literature now extends beyond classical notions of media and information literacy(43) current since the 1970s to embrace data literacy,(44) awareness and understanding of how personal data is used, and security literacy, to mention but two potential areas of expansion that may be relevant for the mass emergence of informed digital citizens. (45) How far to expand into these areas would be a matter for the civil society dialogue process described below. Policy push from Brussels in favour of independent funding of media and information literacy is unlikely to succeed unless it is somehow anchored in the Member States, although the HLEG notes some schemes like the Safer Internet Centres and the Radicalisation Awareness Network which provide proven models for EU intervention, with strong decentralisation to civil society and media actors. Many media organisations have played an honourable role in developing good media literacy practices over the years as detailed in section 2. With the rise of the disinformation phenomenon, platforms have started to play a role in this area. There is however a need to think more strategically about how media literacy is implemented across Europe.(46) The challenges facing far-reaching media literacy projects in Europe include above all the diversity of national education systems; a lack of focus by national curricula on critical media literacy; and, even in countries where these skills are either mandated or supported at government level, low uptake from educators at grassroots level. A lack of cross-border collaboration on best practice, silo approaches dividing academia, civil society organisations, educators and media compounds this challenge. For media and information literacy to be effective, it must be implemented on a massive scale in school curricula and in teacher training curricula, with clear methods of evaluation and cross-country comparison and with reflection in educa(42) Independent research suggest analytic thinking is one factor that influences how receptive people are to disinformation, see e.g. Pennycook, G. & Rand, D. G. (2017) Who Falls for Fake News? The Roles of Analytic Thinking, Motivated Reasoning, Political Ideology, and Bullshit Receptivity. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3023545 JA (43) See Renee Hobbs, Digital and Media Literacy - A Plan of Action: A White Paper on the Digital and Media Literacy Recommendations of the Knight Commission on the Information Needs of Communities in a Democracy, Aspen Institute 2010; also David Buckingham, Defining digital literacy: What do young people need to know about digital media? Nordic Journal of Digital Literacy, 2006–2016, p. 21–34 (44) See Alava, Seraphin, Frau-Meigs, Divina and Hassan Gayda. 2017. Social Media and the Radicalization of youth in the digital era. Paris: UNESCO. (45) See Frau-Meigs, Divina, O’Neil Brian. Tome, Vitor, Soriano Alessandro. 2017. Competences in Digital Citizenship Education. Strasbourg: Council of Europe (46) See Mapping of media literacy practices and actions in EU28, European Audiovisual Observatory, March 2017, https://ec.europa.eu/digital-singlemarket/en/news/reporting-media-literacy-europe One overarching solution to help resolve the issues of fragmentation would be to develop a community of practice across the EU, with the full involvement of civil society across the Member States. This has not occurred on a wide scale before and should be a key part of the approach, to ensure that best practices spread beyond their origins and are accessible in all Member States. For this to happen, cross-border co-operations should be supported with adequate funding. This should be therefore a multi-stakeholder approach with public-private partners, involving the EU, Member States and market actors from the media and platform sectors. In the light of the above considerations, HLEG recommends the following for improving media literacy within the European Union: (i) Actions promoting a reassessment and adjustment of educational policies Key competencies and school rankings across EU. “Media and information literacies” are acknowledged by HLEG as crucial for resisting digital disinformation and should be included explicitly to the list of “EU key competences for lifelong learning”(48) These key competences are expected to serve as a long-standing reference for all future educational and curricula reforms in EU and Member State levels. Inclusion would enable coherent policy formulations and responses in view of rest of HLEG’s recommendations. European institutions and national governments should recognize media and information literacy as core literacy, adding it into school curricula and adding this to the measurements (47) See Frau-Meigs, Divina, Irma Velez, Julieta Flores Michel (eds). 2017. European Public Policies on Media and Information Literacies in Comparative Perspective. London: Routledge, 2017 (48) See https://ec.europa.eu/education/sites/education/files/annex-recommendation-key-competences-lifelong-learning.pdf 4. RESPONSES AND ACTION that determine all school rankings, particularly in the OECD’s PISA rankings. The EU should make this a stated priority with the aim of integrating critical media literacy into the core literacies guaranteed to all schoolchildren in Europe, with formal status in national school curricula. This can engage libraries as well. The HLEG recommends the EU to transmit this recognition beyond Europe via membership of OECD and Unesco. Training for Teachers. European institutions and national governments should mandate teacher training colleges to include critical media literacy modules and encourage critical media literacy to become an integral part of all subject-learning, lifelong learning for teachers. It recommends further that the EU support such teachertraining through Erasmus+, Training and Education 2020 and similar schemes. It recommends further that professional media and fact-checkers commit to playing an active role in media literacy education, adding their field experience and technological expertise to tested and verified media literacy approaches. (ii) Actions in support of media and information literacy programmes for citizens of all ages Efficiency, Best Practice and Evaluation. For media and information literacy efforts to be effective in a rapidly changing field, best practices and evaluation are critical. All organisations currently engaged in developing media literacy practices must work to abandon silos, collaborating across civil society, academia, educational authorities, European institutions and e.g. health professionals. Existing work done on best practice across the EU includes the European Audiovisual Observatory or ET2020 Working Groups. Collaborations already in place include the Safer Internet Centres, Better Internet Centres and Schoolnet.eu. While these currently focus on young people, their model could be used to create Europe-wide initiatives on other age and demographic groups. Such collaborations should aim to maximise efficiencies across EU national borders/languages. Depending on the country’s educational system, non-governmental organisations can play an important, auxiliary role in the system. CSO, working with academia, educational psychology professionals and the media and social media industry, should formulate skill and age-specific media and information literacy approaches and monitor their effectiveness. In so doing, they should focus where possible on adapting language to target audiences on the much needed analytical framework to understand different types of disinformation. This could also include the development of open source tools to tackle disinformation on an everyday basis as a consumer of news. A specific action for citizens of different age groups should be aimed at sensitizing voters to the importance of integrity of elections. The media and information literacy community should join forces and promote literacy programmes to enhance the quality of information around elections and support a fact-based and data-checked public debates. The enhancement of an EU wide media literacy community needs funding for credible, effective and independent media and information literacy actions. Such funding could be channelled via a public-private-civic partnership as a dedicated financing facility. Its task will be, inter alia, to develop, support and evaluate initiatives to combat mis- and disinformation at all levels. Critically, its task will be to shield funds recipients from any real or perceived undue partisan interests and provide full transparency to the public. Regional and networked approaches. The European Commission should encourage and support initiatives that promote media and information literacy to tackle the specific needs of a sub-region (e.g. Baltic or Eastern Europe) and with a concentration on interdisciplinary work between CSOs and media publishers. On a more general note, the HLEG stresses the importance of regular reporting on media and information literacy from Member States and all relevant stakeholders in order to foster evaluation, sharing of practices, monitoring of disinformation efforts, outreach of projects to and to ensure the development of full-fledged communities of practice.(49) c. Empowerment of users and journalists Empowering users of platforms’ services, both citizens and media professionals, is a key element to increase the resilience of society to various forms of disinformation. To achieve this goal it is necessary to develop and make available tools that aggregate quality signals (e.g. identity labels, source transparency indicators, verified content labels) while increasing users’ control over the content they search or are recommended online through the development of personalised content filtering systems aimed at facilitating the display online of the widest possible range of relevant news and information to consumers. There are many different approaches as to how parameters like quality and diversification of content could be taken into account by platform services. Such tools would allow users (49) for more, see the discussion on AVMSD revision. 27 28 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION to better navigate the vast amount of online information while leaving them the possibility to interact within the logics that drive the algorithmic selection of content. Obviously all of these tools and approaches should be fully tested and evaluated to avoid any negative effects. Actions should be taken to foster the creation of a competitive market of such applications and give each user the freedom to select the tools that he/she consider most suitable, taking into account the necessity for interoperability of apps and systems. At the same time, media professionals have to rely more and more on digital content in order to remain at the cuttingedge of information search techniques, as well as audience reach and editorial mix decisions. Due to the fast pace of current technological change, they experience difficulties in adjusting their production processes to a digital ecosystem which is increasingly distributed across a variety of new actors (e.g. citizen journalists, eyewitnesses, bloggers, influencers) and works faster than ever. Journalists need to master technologies that help in discovering breaking news and verifying the veracity of online audiovisual and text material in order not to amplify false information. Building effective fact-checking and good debunking strategies require working with tools that provide them with relevant information in real time about possible false information spreading online, its sources and other relevant contextual elements of trending and viral spread. (ii) Empowerment of journalists In the HLEG opinion, these challenges could be addressed through the following actions. (i) Development of online tools for user empowerment Client-based interfaces should be promoted. This may include the development of built-in tools/plug-ins and applications for browsers and smartphones, to empower users to better control access to digital information. In particular, platforms should consider ways to encourage users’ control over the selection of the content to be displayed as results of a search and/or in news feeds. Such system should give to the user the opportunity to have content displayed according to quality signals. Moreover, content recommendation systems that expose different sources and different viewpoints around trending topics should be made available to users in online platforms. Such system should provide a certain degree of control to users. User empowerment is part of the multi-dimensional approach in this report, it should always coincide with media and information literacy, and the use of the tools should be part of the education. Data on the use should be made available for independent researchers to assess the use and effectiveness of such user empowerment. Platforms should be encouraged to work on strengthening their existing online tools or, if applicable, developing new online tools allowing users to exercise their right to reply and to correct false stories. Improved reporting systems would permit more systematic retracting of incorrect stories. In order to keep their brand reputation and retain customers’ trust, media outlets will have to equip newsrooms with professional automatic content verification tools for audiovisual and text-based reports spread online, preferably as part of an industry wide solution. Investments are also needed in order to create interdisciplinary fact-checking and verification teams that are specialised in identifying disinformation. Voluntary partnerships with social media platforms and external fact checking initiatives are encouraged for cross-fertilization of requirements, knowledge and tools, as well as to share investment costs and maximise their impact. Source checking tools are however equally important instruments to empower journalists. Newsrooms in particular need more powerful tools to be able to visually map online networks and connections to understand how disinformation is being created, spread and amplified. When assessing the credibility of a piece of information, the source who originally created the content or first shared it, can provide further evidence about whether something is accurate. For example, routinely journalists could be researching the date and location embedded in domain registration information of a supposed ‘news site’ to seeing whether it was newly created. Training for journalists (possibly supported by public funding schemes, see Section 4.d below) should also be considered with a view to reducing risks of disinformation making its way into their reporting activities. Particular attention should be given to training modules including media and information literacy, as well as to the role of technology, which can provide a great support for the identification of false information. In choosing projects to be supported, special attention shall be given to those focusing on Central and Eastern Europe countries. Media innovation projects. Funding for news media sector innovation is mainly coming from the private sector. The HLEG encourages platforms to continue projects that empower journalists and extend them to more Member States. Furthermore, media outlets should also invest more in innovation. The HLEG strongly recommends that the European Union increases R&I funding dedicated to news media innovation and more specifically for innovation in dealing with disinformation. Such funding should target fact checking tools, artificial intelligence, augmented newsrooms, conversation journalism, language technologies and big data for media. 4. RESPONSES AND ACTION d. Diversity and sustainability of the news media ecosystem source verification work, and disinformation monitoring, investing in media and information literacy, and funding research, rather than to government initiatives. The HLEG strongly believes that tacking the effects of disinformation without incurring into any form of direct or indirect censorship presupposes a critical, engaged and well-informed readership. Real and quality news’ rate of return is declining and remains by definition lower than disinformation’s rate of return, as the latter is usually cheaper to produce. It is clear therefore that a strategy aimed at “diluting” disinformation through increased transparency and enhanced visibility and findability of trusted news content can only achieve its goals if combined with actions designed to preserve the diversity and long-term economic sustainability of the news media ecosystem. (50) Actions in support of press freedom and pluralism. Public authorities should commit to providing an enabling environment for substantial media pluralism through a combination of de facto and de jure protection of basic rights to free expression and diverse information, including appropriate forms of indirect and potentially direct support for private sector media, and support independent public service media which can help produce quality information, counter disinformation.(51) Together such responses can help enhance media pluralism and maximise diversity in the digital landscape/ environment. This is an area that requires a renewed commitment by public authorities to stand behind and defend fundamental principles, and in particular freedom of expression, free press and media pluralism. This is also an area that requires careful consideration of the challenges that affect the news media industry in the transition from the traditional off-line environment to the online news distribution model. Independent and pluralistic media are the cornerstone of a democratic society and the sustainability of journalism is key to continue to offer high quality content to consumers. Ways should be found to keep (research) journalism financially viable to produce quality content. A serious, constructive effort to train, educate and prepare data journalists is crucial. (i) Actions at European level In the US, the State Department has been given $120 million to counter foreign efforts to meddle in elections or sow distrust in democracy, and the Defence Department has been tasked with dedicating further resources to combating propaganda. The HLEG hopes that the European Commission and EU member states through Horizon 2020, the next MFF, Member State initiatives, and independent funding vehicles will make at least a similar level financial commitment. However, it would recommend that the majority of the funds invested are dedicated to supporting independent quality news media, training journalists, empowering fact-checkers, (50) Research suggests that detailed counter-messages and alternative narratives are often more effective than corrections in countering disinformation. See e.g. Chan, M. P. S., Jones, C. R., Jamieson, K. H. & Albarracín, D. (2017) Debunking: A meta-analysis of the psychological efficacy of messages countering misinformation. Psychological Science. 28 (11), 1531-1546. Beyond this, providing information from trusted sources can help counter disinformation, see e.g. Vraga, E. K. & Bode, L. (2017) I do not believe you: How providing a source corrects health misperceptions across social media platforms. Information, Communication & Society. 1−17. https://doi.org/10.1080/1369118x.2017.131388, https://doi. org/10.1177/0956797617714579 Press freedom could become an issue if black lists of media were to be established. As regards the notion of “journalistic standards”, they should be set by practitioners and the media industry itself, not by others. Funding of projects supporting quality journalism. Professional media needs to continue to invest in quality journalism, including data literacy, data narrative, data visualization and collection. Moreover, more cross-border cooperation between news media organisations may have positive effects on both quality standards and raise collective awareness about false narratives across newsrooms. However, such endeavours are often difficult to fund. The HLEG takes note of the fact that, in addition to actions in support of media freedom and pluralism, the Commission is currently providing funding to media consortia for projects aimed at reporting on EU affairs from a pan-European point of view and at making innovative use of data-driven techniques, while respecting complete editorial independence. (52) While not in any way being detrimental to existing forms of State Aid, media support could include, for example, VAT exemptions or other types of tax breaks. These are areas that the HLEG would urge Member States to examine to ensure a vibrant media sector that underpins healthy democracy in the EU. Investing in research and innovation actions to improve technologies for online media services and for moder(51) Different Member States have different degrees of de facto media pluralism, and many face challenges, especially around market plurality, political independence, and social inclusivity according to the European University Institute’s Media Pluralism Monitor: http://cmpf.eui.eu/mediapluralism-monitor/mpm-2016-results/ (52) In particular, the project “European Data News Hub” involves several news agencies to build an online hub offering free news about EU affairs which offers data-driven news, while the “European Data Journalism Network” (EDJNet) produces, shares and publishes data-driven content on European affairs, available free of charge through a multilingual and open source website. 29 30 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION nization of news rooms. Current innovation projects are i.a. funded through Google DNI and Facebook Journalism Project. The EU needs to step up its role in terms of media innovation. The Horizon 2020 programme – and its successor – should include more specific media-related calls in general, and in particular for innovation in dealing with disinformation. Such calls could include the use of fact checking tools, artificial intelligence, augmented newsrooms, conversation journalism, language technologies and big data for media. Generally, research and innovation in the area of news, media, media effects and content distribution should be supported. The role of researchers and academics is essential to understanding the issues and challenges we are facing, and to forging a pathway forward. By ensuring an inclusive ecosystem approach, researchers and platforms can better coordinate to address these challenges. There is an understandable interest from researchers in securing greater access to information and data on platforms, e.g. through APIs. Platforms are unique in the way they provide access to information depending on their technological design. Therefore, measures to access information will always be reliant on the type of platform.  This should be an area of further work to improve access to data while continuing to protect the users of services especially in respect of requirements related to privacy and data protection, and to avoid any other potential unintended consequences. Journalists, media and legitimate news sources need protection from state and political interference and agenda setting. EU Member States signed in 2016 the Council of Europe Recommendation on the Protection of Journalism. Unfortunately no Member State has implemented it to date. A concerted effort by EU Member States to ensure rapid implementation is needed. Self-regulatory Codes of Practice which set out both professional standards for journalists and rules establishing complaint mechanisms for inaccurate or intrusive reporting should be broadly implemented. These Codes would help generate trust in the quality of news. In this context, and to that extent, platforms should commit to ensure access to data for independent inquiries, audits and research into activities reliant on proprietary media and platform data infrastructures in order to ensure that European citizens are better informed and that responses to problems of disinformation can be evaluated. Platforms, researchers, and academics should work together on appropriate tools and datasets to improve research into disinformation in Europe. Any type of state aid should be set out through horizontal schemes, assorted with adequate conditions and transparently applied by Member States, to respect the principles of free, independent press and media pluralism, as well as free competition. (ii) Actions at national level No interference by public authorities with editorial independence. There is a long tradition of opposition to government regulation of free press. We would have similar concerns about governments gaining excessive control over what news sources can and cannot be accessed online. Solutions requiring government regulators to favour or deprioritise particular news sources are high risk from a freedom of expression point of view and should not be encouraged. The HLEG does not consider governmental control of digital media to be appropriate solutions to digital disinformation. Independence of media is imperative for Europe’s democratic foundations. Protection of fundamental rights. Legal approaches amounting to well-intentioned censorship are neither justified nor efficient for disinformation. Right of defence and speed are incompatible. Public reaction to censorship will backfire, as ‘the establishment’ or ‘parties in power’ could be (mis-)perceived as manipulating the news to their advantage. Public support, including in the form of State Aid. Careful consideration should be given to the use of public funding for well-defined activities aimed at increasing the long-term economic sustainability of a pluralistic news media landscape, while at the same time not be detrimental to existing forms of State Aid. As the European Commission has the exclusive competence to authorise State aid notified by Member States and establish its lawfulness under Article 107 of the EU Treaty, the Commission can provide Member States with relevant guidance for the media sector to increase legal certainty and enhance the predictability of State aid rules enforcement in this area. While not in any way being detrimental to existing forms of State Aid, this could include for example VAT exemptions or other types of tax breaks. The focus could include training of journalists including digital skills, innovation in news media services including cross-border cooperation and exchanges, news outlets with business models nor reliant on advertisement revenues or targeting sub-scale linguistic markets. 4. RESPONSES AND ACTION e. Process and evaluation  (i) A structured implementation framework  Ensuring that the above responses and actions can tackle the issue of disinformation in a meaningful and effective way requires a structured process of implementation and evaluation, including a clear timeframe to calibrate the effectiveness of the responses through continuous research of the impact of disinformation in Europe.  The HLEG believes that, as a first step, the best responses are likely to be those driven by multi-stakeholder collaborations. Regulatory responses may quickly become inadequate to tackle a multi-faceted problem such as disinformation, whose nature and characteristics are bound to change fast with the evolution of technologies and digital behaviour patterns. The HLEG invites therefore the European Commission to consider how it can promote a general, Europeanwide Code of Practices reflecting the respective roles and responsibilities of relevant stakeholders, especially online platforms, media organisations, fact-checking and research organisations.  It should be acknowledged, however, that the ability of selfregulatory measures in this area has never been tested and the willingness of all parties to adhere to such an approach remains to be proven. Moreover, their effective and consistent implementation across the whole EU may represent a challenge for all players concerned. Therefore, based on an independent evaluation of the effectiveness of the proposed general Code of Practices, the European Commission should consider, in a second step, appropriate (co) regulatory responses or competition instruments in order to ensure that the actions recommended in the present Report are effectively implemented. In this connection, it is important to emphasise that regulatory responses need to take account of existing market solutions, ensure due process, as well as accountability and proportionality taking the fundamental rights perspective as a starting point.  Given the long-term scope of some actions, whether or not the responses set out by the HLEG can achieve their intended goals greatly depends on their continuity within the European Commission’s policy agenda. Actions suggested in this report should therefore create a pipeline to further discussion, possible responses and actions. It is however foremost important to ensure an updated, evidence-based understanding of the scale, scope, technologies, tools and precise nature of the problems at hand, as disinformation problems are deeply intertwined with the development of digital media and the evolution of the digital environment, while the tactics and techniques underpinning disinformation strategies will continue to evolve. To ensure appropriate, efficient, and future-proof responses, it will be necessary therefore to continuously examine the problem and the interactions between all relevant stakeholders, which can facilitate further cooperation. The definition of the roles that private and public actors should play in the implementations of the proposed actions should therefore take into account this dynamic.  (ii) Scope of a general European Code of Practices to counter disinformation  As a general aim, the proposed Code of Practices should establish a multi-stakeholder approach (see also sections 4.a and 4.c above). To this end, it needs to:  ▶▶ Clearly identify the target stakeholders and aim for the widest possible uptake across the EU; ▶▶ Set out clear rules for each stakeholder group based on the principles agreed within the HLEG (see below), and provide for their periodical review in light of the progress and efficiency of the self-regulatory process; ▶▶ Establish a mechanism for an independent and permanent evaluation of the measures taken by the parties to ensure its implementation; ▶▶ Ensure coordination with the European Centres for research on disinformation; ▶▶ Take stock of what exists already across all stakeholders, identify gaps and fill these gaps with a particular view on platforms; ▶▶ Not replace existing self-regulatory systems or superimpose additional layers; ▶▶ Consider existing mechanisms of implementation, for example regarding trust and transparency indicators, and synchronize with them.  (iii) A multi-stakeholder engagement process  The proposed Code of Practices should clearly identify the target stakeholders to be part of the engagement process.  Since filtering out disinformation is difficult to achieve without hitting legitimate content, and is therefore problematic from a freedom of expression perspective, it is necessary to improve the findability of, and access to, trustworthy content. Hence, media companies, both print press and broadcasters, have an important role in promoting quality and reliable 31 32 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION information and in debunking misleading or incorrect information. Providing all audiences with independent news, and helping them to better understand society, represent an essential contribution to informed citizenship. This underpins the values of democratic societies, offering a wide choice of quality content, impartial information and pluralistic views, promoting both social cohesion and cultural diversity. In order to foster public trust and citizens’ reliance on the information services provided by media outlets assuming editorial responsibility, it is important for media companies to be part of the proposed multi-stakeholder Code of Practices.  The same considerations apply to nascent organisations engaged in fact-checking and source-checking activities. It is necessary to create the basis for a trusted network of independent researchers and journalists assessing the veracity of factual claims and mapping the sources contributing to the spread of disinformation across different areas of public interest, including general affairs, politics, health, science, education, finance and more. Such organisations should also therefore be part of the joint endeavour to set out general self-regulatory principles for countering disinformation.  Finally, online platforms are a key actor for limiting the spread of disinformation and improving the visibility and findability of trusted news sources. Digital intermediaries such as social networks and online video platforms can impact public opinion by sorting, selecting and ranking news and information via their algorithms. They should therefore be able and willing to act in a responsible way that is commensurate with their powers and the impact that their activities can have on forming public opinion While a number of major online platforms have already taken positive measures as explained in Section 2 above, no clear and binding rules of conduct specifically designed to tackle disinformation online have been established so far. Hence, platforms need to play a central role in setting out and committing to the proposed Code of Practices, to further increase the transparency of their activities as information intermediaries, without delay. Accepting a level of public accountability is a way of building trust for platforms as intermediaries, and to help end-users make better informed choices.  A Coalition representing relevant actors, including online platforms, news media organisations (namely press and broadcasters), journalists, publishers, independent content creators, the advertising industry fact-checkers and other practitioners, should be established. The purpose of the Coalition would be to elaborate, on the basis of the Key Principles below, a multi-stakeholders Code of Practices and ensure its implementation and continuous monitoring and review. A first order of business would be to establish the form of governance for the Coalition. As for recruitment, the process should be open and transparent. The Commission is invited to consider in line with best practices how to support the creation of such Coalition and employ as a sounding board academics, technical experts, civil society organisations and fundamental right, privacy and freedom of expression specialists. (iv) Key principles and roadmap to guide the setting out of the Code of Practices The following key principles have been agreed within the HLEG as a starting point for defining specific rules for the proposed Code of Practice. The elaboration of such a Code will be the task of a Coalition mentioned under point 4.e (iii) above, which will strive to involve all relevant stakeholders from the relevant sectors during the process. The envisaged multi-stakeholder Coalition will be chaired by an expert offering sufficient guarantees of competence and independence. Given the absence of relevant codes of practice applicable to platforms, the HLEG considered necessary to agree on a set of guiding principles which takes due consideration of the range of platforms and therefore the need to carefully tailor solution(53). These principles should serve as a starting point for the development of a Code of Practice within the Coalition. While these guiding principles are to be further tailored-made to reflect the specificities of individual platforms, they should be used to pursue the ongoing discussions with all stakeholders committed to ensuring quality news and information online. Without prejudice to the further work of the Coalition, the ten guiding principles, the Key Principles, supported by the HLEG are the following   1. Platforms should adapt their advertising policies, including adhering to “follow-the-money” principle, whilst preventing incentives that leads to disinformation, such as to discourage the dissemination and amplification of disinformation for profit. These policies must be based on clear, transparent, and non-discriminatory criteria;  2. Platforms should ensure transparency and public accountability with regard to the processing of users’ data for advertisement placements, with due respect to privacy, freedom of expression and media pluralism;  3. 3. Platforms should ensure that sponsored content, including political advertising, is appropriately distinguished from other content;  (53) For the sake of clarity, it should be recalled that the concept of “platform” has been previously defined in footnote 3 4. RESPONSES AND ACTION 4. Platforms should take the necessary measures to enable privacy-compliant access to data for factchecking and research activities;  5. Platforms should make available to their users advanced settings and controls to empower them to customise their online experience;  6. Platforms should, in cooperation with public and private European news outlets, where appropriate take effective measures to improve the visibility of reliable, trustworthy news and facilitate users’ access to it;  7. Where appropriate, trending news items should, if technically feasible, be accompanied by related news suggestions;  ▶▶ As a first step, the Coalition should start as soon as possible the elaboration of the Code of Practices and, in parallel, the should • agree on key performance indicators (KPIs) which should be included in the Code and be used to evaluate the effectiveness of the Code based on the results of its implementation, with the situation at 1 January 2019 being taken as baseline scenario; • further reflect on the definition of adequate source transparency indicators which should help the effective implementation of the transparency measures required from platforms; The above steps should be completed by July 2018 8. Platforms should, where appropriate, provide userfriendly tools to enable users to link up with trusted fact-checking sources and allow users to exercise their right to reply;  ▶▶ In July 2018, the European Commission should commission an independent Progress Assessment Report (PAR) to provide expert advice on the adequacy of the provisions set out in the Code, the robustness of the agreed KPIs and its uptake by stakeholders willing to undersign and able to implement the Code; 9. Platforms that apply flagging and trust systems that rely on users should design safeguards against their abuse by users;  ▶▶ By end September 2018, the Coalition should record the expressions of intent provided by stakeholders willing to undersign and able to implement the Code; 10. Platforms should cooperate by i.a. providing relevant data on the functioning of their services including data for independent investigation by academic researchers and general information on algorithms in order to find a common approach to address the dissemination and amplification of disinformation. ▶▶ In October/November 2018, the PAR should be made public and submitted to the HLEG; As regards media organisations, it is proposed that the Code refer to the existing journalism deontological codes, ethics and standards to guarantee quality in the methods in which news is produced, including guidelines issued by international organisations such as the International Federation of Journalists or national bodies. It is also suggested that, in this context, the role played by Press Councils and their European organisation, the Alliance of Independent Press Councils of Europe (AIPCE), should be highlighted, including as regards the implementation of source transparency indicators.  As regards fact-checking organisations, the proposed Code could take the existing IFCN Code of Principles as a valid reference.  In order to roll out the initiative for the Code of Practices the HLEG agrees on the following steps and recalls that they reflect an intense debate across stakeholders, which started with the first meeting of the HLEG on 15 January 2018:  ▶▶ In November 2018, the HLEG will consider the matter based on both the work of the Coalition and the PAR, and formulate recommendations, including possible adjustments, accompanying measures for further monitoring and evaluation to ensure compliance with the Code of Practices, or options for appropriate regulatory interventions or additional policy initiatives; ▶▶ 1 January 2019, the members of the Coalition that are signatories of the Code start its implementation; ▶▶ March 2019, ahead of the EU Parliament elections and taking stock of the progress made and the HLEG’s advice, the European Commission will issue a new Communication examining the options available This may include additional fact-finding and/or additional policy initiatives, using any relevant instrument, including competition instruments or other mechanisms to ensure continuous monitoring and evaluation of the implementation of the Code, see also below (v).  ▶▶ 1 July 2019: second PAR based on the validated KPIs and review by a HLEG. 33 34 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION (v) A process for an independent and permanent evaluation  In case the initial Progress Assessment Report and the HLEG considers appropriate to pursue the self-regulatory efforts mentioned above, the European Commission should specify, by April 2019, in which way it intends to ensure the continuous effectiveness of the Code.  To this effect, it is recommended that the Commission establishes a permanent review mechanism entrusted with an independent expert body which, on the basis of the agreed KPIs, will carry out an in-depth analysis of the progress made in countering disinformation through the Code of Practices, as well as through the implementation of the other recommendations formulated by the HLEG which, as described in sections 4.b and 4.d above, include actions aimed at promoting media and information literacy and those in support of the diversity and sustainability of the news media ecosystem.  Such a mechanism should entail periodical reviews and include elements of continuity, accountability and transparency factors such as progress reports, communications and timeframes. Taking into account the outcome of such periodical reviews, the Commission should decide the appropriate course for further action. (vi) Coordination with the European Centres for Research on Disinformation  The HLEG considers that the effectiveness of the proposed responses is strongly dependent on the development of the European Centres for interdisciplinary and independent evidence-based research on problems of disinformation mentioned Section 4.a above.  In its upcoming Communication, the Commission should clarify how to ensure continuity for the proposed engagement process by, for instance, addressing the need for a Europeanled initiative aimed at stepping up transparency efforts by boosting fact- and source checking activities and continuous research in the area of disinformation on digital media. The proposed creation of a network of European Centres for Research on Disinformation across European countries should involve the participation of independent national research organisations, which should manage multidisciplinary teams linking fact-checking organisations, scientists, members of press councils and open to collaborations with news media, advertisers’ representatives and platforms. Such teams could take the form of “living labs” following the example of existing support schemes for R&D&I activities successfully implemented in several Member States. While support to such national entities could be provided by Member States in combination with private contributions, the European Commission should assure a coordinating role (for instance by creating a Centre of Excellence) and ensure the necessary infrastructure for effective networking which could be supported along the example of other digital service infrastructures currently in operation. In this respect, it is important that long-term funding will be secured within the next multiannual financial framework (MFF) of the EU. 5. CONCLUSIONS: SUMMARY OF ACTIONS 5. Conclusions: summary of actions The analysis presented in this Report starts from a shared understanding of disinformation as a phenomenon that goes well beyond the restrictive and misleading term “fake news”. Disinformation as defined in this Report includes all forms of false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit. It does not cover issues arising from the creation and dissemination online of illegal content (notably defamation, hate speech, incitement to violence), which are subject to regulatory remedies under EU or national laws, nor other forms of deliberate but not misleading distortions of facts such a satire and parody. It acknowledges that, while not necessarily illegal, disinformation can nonetheless be harmful for citizens and society at large. The risk of harm includes threats to democratic political processes and to democratic values that shape public policies in a variety of sectors, such as health, science, education, finance and more. It is driven by the production and promotion of disinformation for economic gains or for political or ideological goals, and can be exacerbated by how different audiences and communities receive, engage, and amplify disinformation. Moving from this premise, The HLEG has identified a number of interconnected and mutually reinforcing responses that take account of the fact that various forms of disinformation, and the threats they pose to society, are deeply intertwined with the fast development of digital media, while the tactics and techniques used to amplify its spread online will continue to evolve. Therefore, in order to ensure appropriate responses, which are fully compliant with the fundamental principles of freedom of expression, free press and pluralism, and at the same time are future-proof and efficient in averting public harm, the HLEG cautions against simplistic solutions. It recommends instead a multi-dimensional approach that caters for the need to continually examine the phenomenon and evaluate the effectiveness of the concrete measures adopted by different actors. The advice of the HLEG rests on five pillars that, as a set of inter-dependent actions, form part of its overarching, multi-dimensional approach. These five pillars consist of actions designed to: a. enhance transparency of the digital information ecosystem, b. promote and sharpen the use of media and information literacy approaches to counter disinformation and help users navigate our digital information environment, c. develop tools for empowering users and journalists and foster a positive engagement with fast-evolving information technologies, d. safeguard the diversity and sustainability of the European news media ecosystem, and e. calibrate the effectiveness of the responses through continuous research on the impact of disinformation in Europe and an engagement process that includes predefined and time-framed steps combined with monitoring and reporting requirements. In the light of the above, the HLEG calls on the European Commission to consider, in its upcoming Communication on fake news and online disinformation, the proposed multi-dimensional approach, whose main features can be summarised as follows. From a general policy perspective, the HLEG suggests, as a first step, a self-regulatory approach based on a clearly defined multi-stakeholder engagement process, framed within a binding roadmap for implementation, and focused on a set of short and medium-term actions. This should go in parallel to the proposed interventions to strengthen media and information literacy and the diversity and sustainability of the digital information ecosystem, actions that by their own very nature have a longer time horizon. In a second step, an intermediate evaluation of the effectiveness and efficiency of these short and medium-term measures should then lead the Commission to re-examine the matter in Spring 2019, with a view to deciding whether further measures, including (co)regulatory interventions, competition instruments or mechanisms to ensure a continuous monitoring and evaluation of self-regulatory measures, should be considered for the next European Commission term. 35 36 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION a. Recommendations for the short to medium-term The following recommendations are designed to foster transparency, algorithm accountability and trust-enhancing practices, while contributing to empower users and journalists, in line with the responses set out in Section 4.a and 4.c of this Report. They are based on a multi-stakeholders approach that recognises the need for cross-sector and cross-border cooperation as a precondition for tackling disinformation, a multi-faceted problem that does not have one single root cause and cannot therefore be addressed by a single measure. • active participation within the proposed European Centres for research on disinformation to facilitate access to data within a safe and privacy-compliant space, so as to foster a better understanding of the working of algorithms and of the mechanisms that enable the spread of disinformation online. The aim of such collaboration is to enhance research, and define and adjust appropriate responses that take into account national or regional specificities and are proportionate to the risk of public harm likely to be caused. The role of public authorities i. The European Commission should: The role of platforms, news media and factchecking organisations • Facilitate the creation of a multi-stakeholder Coalition against disinformation and assist stakeholders in developing a Code of Practices and in evaluating its effectiveness. i. Establishment of a Coalition representing online platforms, news media organisations (including press and broadcasters) and civil society organisations with expertise in fact-checking, which will strive to involve all willing stakeholders from the relevant sectors during the process. Its main task will be to ensure the elaboration of the proposed multi-stakeholders Code of Practices and accompany its implementation and continuous monitoring. • In light of the Progress Assessment Report mentioned above, and by March 2019, consider options that may include additional fact-finding and/or additional policy initiatives, using any relevant instrument, such as competition instruments or mechanisms to ensure a continuous monitoring and evaluation of self-regulatory measures. ii. Definition of a multi-stakeholder Code of Practices setting out the concrete rules of conduct in function of the role which platforms, news media and fact-checking organisations have to play in order to protect an enabling environment for freedom of expression while fostering the transparency and intelligibility of different types of digital information channels. The Code should be built on the Key Principles set out in Section 4.e of this Report and provide for a binding Roadmap for implementation, including an initial Progress Assessment to be carried out by an independent expert entity by October/November 2018. The proposed Code is designed to cater for the actions examined in Sections 4.a and 4.c and ensure: • that the Key Principles are made operational in such a manner that the actions formulated in this Report are fully achieved; • a first elaboration and refinement of fair, objective and multi-dimensional source transparency indicators to improve public knowledge about online information sources, transparency of journalistic processes and findability of trustworthy content; • an effective cooperation to increase the transparency and efficiency of fact-and source checking practices to facilitate the dissemination of verified and reliable information; • Support the establishment of European Centres for research on disinformation (see Section 4.a (iii)), operating at national level and aimed at mapping the digital ecosystem for disinformation, its dominant technologies, tools and practices, monitoring the veracity of information through the most advanced fact-checking tools, artificial intelligence, language technologies and big data for media. In particular, the European Commission should provide adequate funding and organise a Centre of Excellence equipped with the necessary service infrastructure enabling an efficient network of such national Centres. • Pursuing and intensifying efforts in support of media innovation projects, including through funding for R&I, to empower journalists in dealing with disinformation. The Commission is invited to tender by summer 2018 an independent study on Media sustainability to inform its policy and budget pipeline for 2019-2024. ii. Member States should: • Facilitate the operation of the European Centres for research on disinformation by providing funding to research organisations that operate innovation hubs or living labs open to fact-checkers, accredited journalists and researchers from different relevant fields, as well as representatives of Press Councils and platforms. 5.CONCLUSIONS: SUMMARY OF ACTIONS b. Recommendations for the longer term The following recommendations are designed to improve media and information literacy in Europe and to support the diversity and sustainability of the news media ecosystem, in line with the responses set out in Section 4.b and 4.d of the present Report. They chiefly call on public authorities, both at national and EU level, to step up their efforts, but they also require the cooperation of civil society organisations , media organisations and platforms. i. The European Commission should: • Sharpen actions in support of media and information literacy for all citizens, including exchange of best practices and training for teachers (e.g. through Erasmus+, Training and Education 2020 and similar schemes), and the promotion of media literacy in EU curricula reforms and OECD PISA competency rankings. Special attention should be paid to the specific needs of certain sub-regions (e.g. Baltic or Eastern Europe); • Following the example of collaborations already in place and currently focusing on young people (e.g. the Safer Internet Centres, Better Internet Centres and Schoolnet.eu), consider increasing its support to build a Europe-wide community of practice engaged in leading media and information initiatives on different age and demographic groups. Such funding could be channelled via a public-private-civic partnership as a dedicated financing facility in such a manner as to fully respect funds recipients’ independence; • Increase funding in support for quality journalism, including cross-border collaborations and data-driven journalism; • Ensure that sufficient resources within the current Horizon 2020 and future programmes are made available to promote research and innovation actions aimed at improving technologies for online media services and for the modernization of newsrooms; • As the European Commission has the exclusive competence to authorise State aid notified by Member States and establish its lawfulness under Article 107 of the EU Treaty, the Commission can provide Member States with relevant guidance for the media sector to increase legal certainty and enhance the predictability of State aid rules enforcement in this area. While not in any way being detrimental to existing forms of State Aid, this could include for example VAT exemptions or other types of tax breaks. Such guidance should clarify the objective and transparent conditions under which aid schemes supporting, for instance, training of journalists, innovation in news media services, cross-border cooperation or the creation of centres for media innovation and journalistic incubation, could be designed without compromising the independence of journalists and news media or unduly distorting competition. ii. Member States should: • Sharpen actions in support of media and information literacy for all citizens, including integration of media and information literacy in teacher training and national curricula requirements; • Refrain from interfering with the editorial independence of media and strive for a rapid implementation of the 2016 Council of Europe Recommendation on the Protection of Journalism; • Step up public funding, consistent with EU State aid rules, of activities to improve the long-term sustainability of a pluralistic news media landscape. This could include revising existing forms of generally available support, including via VAT exemptions or other types of tax breaks, and establishing new and well-targeted forms of aid for news production. The focus of such support could include training of journalists including digital skills, innovation in news media services including cross-border cooperation and exchanges, news outlets with business models nor reliant on advertisement revenues or targeting sub-scale linguistic markets. iii. Civil society organisations should: • Work with academia, educational psychology professionals and the media industry to formulate skill and age-specific media and information literacy approaches and monitor their effectiveness. In so doing, CSOs and academia should focus where possible on adapting language to target audiences and on developing analytical frameworks to understand different types of disinformation; • In collaboration with industry, facilitate the development of open source tools to tackle disinformation on an everyday basis. • Design specific actions for citizens of different age groups aimed at sensitizing voters to the importance of integrity of elections; promote literacy programmes to enhance the quality of information around elections; and support a fact-based and data-checked public debates. • Step up efforts to build a community of practice linking up different media literacy organisations and communities at European level. 37 38 A MULTI-DIMENSIONAL APPROACH TO DISINFORMATION iv. Platforms should: • Develop tools to share standard information sheets to users developed by independent (educational) institutions within media and information literacy programs, raising awareness of digital disinformation and emerging findings about digital risks; v. News media organisations should: • Cooperate with CSOs and academia to formulate and implement skill and age-specific media and information literacy approaches, and for all ages, while pursuing their media literacy projects in cooperation with schools and other educational institutions that target younger generations; • Subject to funding, notably from outside sources, continue investing in quality journalisms and equip newsrooms with professional automatic content verification tools for audio-visual and text-based reports spread online; • Ensure the highest levels of compliance with ethical and professional standards to sustain a pluralistic and trustworthy news media ecosystem. ANNEX I ANNEX I List of the 39 members of the High Level Expert Group on fake news chaired by Professor Madeleine De Cock Buning: Last Name First Name Organisation Allan Richard Facebook Bargaoanu Alina Academic professor / personal capacity Bechmann Anja Academic professor / personal capacity Curran Noel European Broadcasters Union Dimitrov Dimitar Wikimedia Dzsinich Gergely Cybersecurity and cybercrime Advisors Network - Cyan Frau-Meigs Divina Academic professor / personal capacity Fubini Federico Gniffke Kai Goyens Monique Journalist / personal capacity Arbeitsgemeinschaft der öffentlich-rechtlichen Rundfunkanstalten der Bundesrepublik Deutschland - ARD The European Consumer Organisation - BEUC Gutierrez Velazquez Ricardo European Federation for Journalists Jiménez Cruz Clara Journalist / personal capacity Leclercq Christophe EurActiv Lemarchand Grégoire Agence France Presse - AFP Lundblad Nicklas Google MacDonald Raegan Mozilla Firefox Mantzarlis Alexios International Fact-Checking Network – IFCN Poynter Markovski Veni Internet expert / personal capacity Nielsen Rasmus Kleis Academic professor / personal capacity Nieri Gina Mediaset Niklewicz Konrad Civic Institute Polák Juraj Rozhlas a televízia Slovenska - RTVS Pollicino Oreste Professor / personal capacity Raag Ilmar Media Executive / personal capacity Rae Independent News and Media Steenfadt Stephen Giovanni Anda Mikko Sonja Olaf Stjärne Hanna Sveriges Television, represents Nordic Public Service Broadcasting Sundermann Bertelsmann Turner Marc Ziga Stephen Vaisbrodė Neringa Communication adviser / personal capacity Van Wijk Wout News Media Europe Von Reppert-Bismarck Juliane Lie Detectors Riotta Rozukalne Salo Schwetje Turk Academic professor and Journalist / personal capacity Academic professor / personal capacity Faktabaari RTL Group Reporters Sans Frontières Academic professor / private capacity Twitter 39 This report was adopted by the High Level Expert Group on fake news on the 7th of March 2018 with the support of 38 members out of 39. Reporters sans Frontières – RSF voted in favour of the report with the following motivation: RSF supports most parts of the report, for example the description of the problem and the underlying principles and freedoms at stake. As regards the recommendations, however, the composition of the HLEG and also procedural limitations, namely the consensual approach, left little room for dissenting opinions. RSF remains firmly opposed to the approach of a ‘Coalition’ as long as its mandate, the appointment procedures and governance mechanisms remain unclear. Given the sensitivity of the matters at stake, such a supra-national entity should be launched only with robust safeguards against undue influence in place. The European Consumer Organisation – BEUC voted against the report with the following statement BEUC votes against the report because we consider that consumer exposure to disinformation needs to be addressed primarily at its source. The absence from the report and its recommendations of any reference to one of the major potential sources of disinformation – clickbaiting – is unacceptable. Assessing the link between advertising revenue policies of platforms and dissemination of disinformation, namely via a sector inquiry, is from BEUC’s perspective a crucial element to find the appropriate mechanisms to tackle this phenomenon. We hope that at a later stage, the Commission will review its intervention on this basis. 3“NEI' [ZZ-8 [' l0-)i)i I Publications Office ISBN 978-92-79-80419—9