ATTACKS IN LONDON AND MANCHESTER MARCH-JUNE 2017 INDEPENDENT ASSESSMENT OF MI5 AND POLICE INTERNAL REVIEWS UNCLASSIFIED by DAVID ANDERSON Q.C. DECEMBER 2017 © David Anderson 2017 The text of this document (this excludes, where present, the Royal Arms and all departmental or agency logos) may be reproduced free of charge in any format or medium provided that it is reproduced accurately and not in a misleading context. The material must be acknowledged as David Anderson copyright and the document title specified. Where third party material has been identified, permission from the respective copyright holder must be sought. Any enquiries regarding this publication should be sent by post to David Anderson Q.C. at Brick Court Chambers, 7-8 Essex Street, London WC2R 3LD, or via his website www.daqc.co.uk. Print ISBN [978-1-78655-599-1] Web ISBN [978-1-78655-598-4] 12/17 FOREWORD Between March and June 2017, London and Manchester saw four attacks in which men used vehicles, knives and explosives to kill and maim members of the public. Tourists, revellers, worshippers, a police officer and children attending a concert were all targeted. A total of 36 innocent people were killed in the attacks, and almost 200 more were injured. The public response – like that of the police who reacted promptly to each attack – was impressive. Manchester came together in a moving demonstration of solidarity. The international headline which described London as “reeling” missed the mark: resilient, or resolute, would have been closer. Rather than divide the country, these shocking crimes united decent people of all races and religions in sympathy for the victims and condemnation for the attackers. Post-attack hate crimes, unacceptable though they are, cannot obscure this greater truth. Even the most futile acts of terrorism destroy lives, and must where possible be prevented. The fundamentals are sound in the UK, as shown by the 20 plots thwarted over the past four years, seven of them since the Westminster attack in March. Some completed attacks must unfortunately be expected. But after four such incidents over a short period, unsparing reflection was required. The internal reviews that the Home Secretary has asked me to assess have been a timely and productive vehicle for that process. I commend the leadership of both MI5 and CT Policing for their willingness to question the way things have been done in the past. To guard against any faltering of their resolve, I embedded myself for part of every week in Thames House and New Scotland Yard, where I attended internal meetings, reviewed drafts, teased out detail, challenged assumptions, called out complacency, drew attention to omissions, arbitrated differences and occasionally counselled greater boldness. To act as a gadfly on the hide of the beast is not the same as to direct a fully independent review. Yet for organisations which have the ability to be self-critical, internal reviews can be of great value. That was certainly the case here. Because they generated their own recommendations, MI5 and the police have every incentive to see them through. With the necessary commitment of time and resources, there is no reason why this should not be done. The nine classified reports that I have assessed contain detailed and accurate accounts of intelligence-handling prior to each of the four attacks, and a series of pertinent recommendations for operational improvement. They will not remove the risk of terrorist attack: to do so would be manifestly impossible in a free society. But if properly given effect they will strengthen the hand of police and MI5, without compromising the sound legal and ethical framework within which they willingly work. I welcome the care and the energy with which the reviews were conducted, their recommendations, and this opportunity to offer a flavour of them to a wider public. DAVID ANDERSON Q.C. Brick Court Chambers 2 November 2017 EXECUTIVE SUMMARY THE REVIEWS  This report, finalised on 2 November 2017, is an unclassified independent assessment of nine classified internal reviews, stretching to 1150 pages and conducted by MI5 and Counter-Terrorism Policing.  The reviews give a comprehensive account of the handling of intelligence prior to the attacks of March-June 2017 at Westminster, Manchester, London Bridge and Finsbury Park, and of operational improvements identified by MI5 and the police.  I was able to comment on and to influence the reviews throughout, thanks to exemplary cooperation and disclosure by both CT Policing and MI5. (Chapter 4)  The reviews followed appropriate processes, given the time constraints, and were conducted in a careful and trustworthy manner. (5.7-5.13) THE ATTACKS  MI5 and CT Policing have thwarted 20 Islamist terrorist plots in the past four years, resulting in 10 life sentences from the seven plots that have so far come to trial. But in a free society and against a worsening threat background, it is not realistic to expect everything to be stopped. (Chapter 1 and 5.20-5.25)  Three of the six attackers were on MI5’s radar, either as an active subject of interest (Khuram Butt - London Bridge) or as closed subjects of interest (Khalid Masood – Westminster, Salman Abedi - Manchester). (Chapter 2)  It is not the purpose of the internal reviews, or of this report, to cast or apportion blame. But though investigative actions were for the most part sound, many learning points have emerged (3.12-3.24, 3.35-3.47). It is conceivable that the Manchester attack in particular might have been averted had the cards fallen differently: 2.37-2.38, 3.14-3.17 and 5.17. THE FUTURE  MI5 and CT Policing have pushed each other, and persuaded themselves, to contemplate significant change. Their 126 recommendations include commitments to better data exploitation, to wider sharing of information derived from MI5 intelligence (including with neighbourhood policing) and to the consistent assessment and investigation of all terrorist threats, regardless of ideology. (Chapter 3).  Some of the recommendations will have implications for oversight, which the new Investigatory Powers Commissioner’s Office is well equipped to provide. (4.9(a), 5.6)  I endorse, so far as I am qualified to do so, the conclusions and recommendations of the internal reviews. But a meaningful verdict on some of them would require a degree of operational expertise to which I do not lay claim. (5.15-5.19)  If properly implemented, I consider that the recommendations taken as a whole will strengthen MI5 and the police in their ability to stop most terrorist attacks. (5.20-5.29) CONTENTS Page 1. INTRODUCTION 1 2. PRE-ATTACK INTELLIGENCE 10 3. INTERNAL REVIEWS 24 4. INDEPENDENT ASSESSMENT 36 5. CONCLUSION 42 ANNEXES ANNEX 1: LIST OF ACRONYMS 50 ANNEX 2: RECENT CONVICTIONS FOR ISLAMIST ATTACK PLOTS 51 ANNEX 3: OIR TERMS OF REFERENCE 52 ANNEX 4: MY LETTER OF APPOINTMENT 55 ANNEX 5: MI5 INVESTIGATIVE PROCESSES 57 1. INTRODUCTION The attacks under review 1.1 In the spring and summer of 2017 London and Manchester suffered four attacks, killing 41 people (including five of the six attackers) and leaving almost 200 more in hospital. The essential facts of those attacks were as follows: a) Westminster (6 deaths including the attacker): On the afternoon of Wednesday 22 March, 52-year old British-born Khalid Masood drove a Hyundai Tucson SUV, hired in Birmingham a few days earlier, into pedestrians who were crossing Westminster Bridge, London. Three were killed at the time and 32 were admitted to hospital, where one died later and several others were treated for life-changing injuries. Masood then took two carving knives out of the vehicle and fatally stabbed PC Keith Palmer where he was on duty outside the Houses of Parliament. Masood was shot by armed police and died of his injuries. The entire incident lasted around 90 seconds. b) Manchester (23 deaths including the attacker): On the evening of Monday 22 May, 22-year-old British-born Salman Abedi detonated an explosive charge in the foyer of the Manchester Arena, at the end of a concert attended by thousands of children. Abedi was killed in the explosion along with 22 innocent people, 10 of them aged under 20 and the youngest, Saffie Roussos, a girl of 8. A further 116 people required hospital treatment. c) London Bridge (11 deaths including the 3 attackers): On the evening of Saturday 3 June, three men (27-year-old Briton Khuram Butt, 30-year old Moroccan Rachid Redouane and 22-year old Italian/Moroccan Youssef Zaghba) drove a Renault Master van, locally hired earlier that day, into pedestrians on London Bridge, killing two people. Abandoning unused a store of Molotov cocktails and wearing dummy suicide vests, they then left the van armed with large knives, which they used on an apparently random basis to kill six more people in nearby Borough Market and in the vicinity of Borough High Street. Armed police arrived within eight minutes and shot them dead. A total of 11 people were killed, and 45 required hospital treatment. d) Finsbury Park (1 death): Shortly after midnight on Monday 19 June, 47-yearold British-born Darren Osborne drove a Citroen Relay van, hired in Cardiff some days previously, into a crowd of worshippers outside the Finsbury Park Islamic Centre in London. Makram Ali, who had been taken ill and was lying on the ground, was struck by the vehicle and died soon afterwards. 10 other 1 people received hospital treatment for injuries. Osborne was arrested and has been charged with murder.1 Each of these four attacks was speedily identified by Government, police and media as an act of terrorism, and referred to as such in my letter of appointment (Annex 4). I adopt in this report the same terminology, without seeking to influence any court or other body that may subsequently be called on to decide whether it is legally appropriate in any given case. The attacks in context 1.2 The attacks under review were the most deadly terrorist attacks on British soil since the 7/7 London tube and bus bombings of July 2005.2 All four were shocking for their savagery and callousness. The impact of the first three attacks was increased by the fact that they came at the end of a long period in which Islamist terrorism3 had taken multiple lives in neighbouring countries such as France, Belgium and Germany but had not enjoyed equivalent success in Britain. 1.3 The plots were part of an increasingly familiar pattern of Islamist and (to a lesser extent) anti-Muslim terrorist attacks in western countries, including in particular northern Europe. The following points provide context, and an indication that lessons learned from these incidents are likely to be transferrable. 1.4 1 2 3 4 First, the threat level in the UK from so-called “international terrorism” (in practice, Islamist terrorism whether generated at home or abroad) has been assessed by the Joint Terrorism Analysis Centre (JTAC) as SEVERE since August 2014, indicating that Islamist terrorist attacks in the UK are “highly likely”.4 Commentators with access to the relevant intelligence have always been clear The murder charge is in no way indicative of bias in the criminal justice system, as is occasionally suggested on the wilder fringes of public debate. Counts of murder or attempted murder are on the indictment when all the most serious plots are brought to justice, including Islamist plots such as the 21/7 attacks of 2005 and the killing of Lee Rigby in 2013. Though specialist terrorist offences exist, they mostly concern precursor or peripheral activity such as the preparation of terrorist acts or the dissemination of terrorist propaganda. The 7/7 bombings accounted for their four perpetrators and 52 innocent victims, injuring over 700 more. Later successful terrorist attacks in England were targeted on individuals (Mohammed Saleem and Lee Rigby in 2013; Jo Cox MP in 2016). Britons killed by terrorists overseas included the 30 victims of a firearms attack at the beach resort of Sousse, Tunisia, in June 2015. This term is a common way of describing terrorist acts committed (to the bafflement and disgust of most Muslims) in the name of Islam. Islamists aspire to a state ordered in accordance with sharia law. Most Islamists support only peaceful means to that end; but an ideology built partly on their beliefs, influenced by Salafism and promoted by organisations such as al-Qaida and Daesh (ISIS, or the so-called Islamic State), is often used to justify acts of terrorism. The threat level stood at SEVERE for most of the period 2006-2011. It was raised to CRITICAL (an attack is imminent) for a few days after the Manchester attack, until MI5 and police were satisfied that other plotters were not still at large. It was again raised to CRITICAL in the immediate aftermath of the attack (not one of those under review) on a tube train at Parson’s Green in September 2017. 2 that this assessment is realistic.5 They have pointed also to the smaller but still deadly threat from extreme right wing (XRW) terrorism,6 exemplified by the murder of Jo Cox MP in June 2016 and by the proscription of the neo-Nazi group National Action in December 2016.7 1.5 Secondly, the growing scale of the threat from Islamist terrorism is striking. The Director General of MI5, Andrew Parker, spoke in October 2017 of “a dramatic upshift in the threat this year” to “the highest tempo I’ve seen in my 34 year career”.8 Though deaths from Islamist terrorism occur overwhelmingly in Africa, the Middle East and South Asia, the threat has grown recently across the western world, and has been described as “especially diffuse and diverse in the UK”.9 It remains to be seen how this trend will be affected, for good or ill, by the physical collapse of the so-called Islamic State in Syria and Iraq. 1.6 Thirdly, the profiles of the attackers named in 1.1(a)-(c) above display many familiar features. Comparing the five perpetrators of the Westminster, Manchester and London Bridge attacks with those responsible for the 269 Islamist-related terrorist offences in the UK between 1998-2015, as analysed by Hannah Stuart (“the total”):10 (a) All were male, like 93% of the total. 5 6 7 8 9 10 See e.g. the assessment founded on classified materials in my last annual report as Independent Reviewer of Terrorism Legislation, The Terrorism Acts in 2015, December 2016, 2.5-2.10. For the purposes of the present report I was able to inspect recent editions of the weekly G Branch Grid, which takes its name from the department of MI5 concerned with so-called international counter-terrorism, and details the top-priority investigations. G Branch grids from regional offices are also prepared. The volume of credible attackplanning identified in recent editions of the G Branch Grid made them a sobering read, despite my knowledge (acquired as Independent Reviewer of Terrorism Legislation) of the terrorist threat to the UK and UK interests abroad prior to March 2017. People with XRW views, whether violent or not, subscribe to a variety of ideologies ranging from a desire to remove Muslims from a “Christian” country to paganism and full-blown neo-Nazism. Like their Islamist counterparts, they may be brought to the point of otherwise unthinkable atrocities by a mindset that dehumanises the enemy group. David Anderson, The Terrorism Acts in 2015, December 2016, 2.21-2.28. In addition terrorism and paramilitary violence remain facts of life in Northern Ireland, where 24 deaths attributable to the security situation were recorded in the 10 years to March 2017: Police Service of Northern Ireland, Police Recorded Security Situation Statistics, October 2017, Annex 1. Andrew Parker cited figures which bear this out: a record 379 terrorism-related arrests in the year to June 2017, and seven thwarted plots since March 2017, as well as the five which were activated (the four under review, plus the attempt to detonate a bomb on a tube train at Parson’s Green, London, in September 2017). He also recorded that MI5 was currently running “well over 500 live operations involving around 3000 individuals known to be currently involved in extremist activity in some way”, and referred to “the growing pool of over 20,000 individuals that we have looked at in the past in our terrorism investigations” (as to which, see further at 1.26 below). Within its priority casework (as to which, see 1.22 below), MI5 saw more instances of individuals potentially involved in attack-planning in the first half of 2017 than throughout the entirety of 2016. Robin Simcox, European Islamist Plots and Attacks since 2014 – and How the US Can Help Prevent Them, The Heritage Foundation Backgrounder, 1 August 2017. The 23 UK plots (as defined by Simcox) between January 2014 and May 2017 were exceeded by the 51 in France and the 24 in Germany, as well as (per capita) by the 12 in Belgium: ibid. Hannah Stuart, Islamist Terrorism: analysis of offences and attacks in the UK (1998-2015), Henry Jackson Society, 2017. In view of pending legal proceedings, no equivalent material is cited in relation to the profile of Darren Osborne, the Finsbury Park attacker. 3 (b) Three were British (Masood, Abedi, Butt), like 72% of the total. (c) One was a convert to Islam (Masood), like 16% of the total.11 (d) Three resided in London (43% of the total) and one in North West England (10% of the total). (e) Three (Masood, and to a more limited extent Abedi and Butt) were known to the police, like 38% of the total. (f) The same three were known to MI5, like 48% of the total.12 (g) At least one (Butt) had direct links to a proscribed terrorist organisation, as had 44% of the total. His links, in common with 56% of the total who had links with such organisations, were with Al-Muhajiroun (ALM).13 In view of their possible pending trials I say nothing of Hashem Abedi, currently detained in Libya in connection with the Manchester attack, or of the Finsbury Park attacker Darren Osborne who (like Khalid Masood at Westminster) is not alleged to have had accomplices. 11 12 13 14 15 1.7 Fourthly, though the targets of the first three attacks did not extend to the whole of the current range,14 they had strong similarities to the targets of other recent western attacks: political centres (e.g. Oslo 2011, Ottawa 2014, Brussels 2016); concert-goers, revellers and crowds (e.g. Orlando 2016, Paris 2016, Barcelona 2017); and police officers (e.g. Melbourne 2014, Berlin 2015, Charleroi 2016). There are precedents also for attacks on observant Muslims which have crossed the boundary from hate crime to terrorism, including the killing of Mohammed Saleem in the West Midlands in 2013.15 1.8 Fifthly, the modus operandi (MO) of terrorist attacks has diversified and simplified over the years, as Daesh has employed its formidable propaganda effort to inspire rather than to direct acts of terrorism in the west. The attacks under review were typical in style for their time and place: This exceeds by more than four times the proportion of converts in the Muslim population: Simcox, op. cit. Khalid Masood converted in prison, like five other offenders forming part of the total. Stuart, op. cit., viii-xiii. 25% of the total were thus linked with ALM, as against 10% with al-Qaida and 5% with Daesh. The attacks under review did not target Jews, military or political figures, sporting events or critical infrastructure such as mass transit systems. To constitute terrorism, an act must not only be politically, religiously, racially or ideologically motivated, but be intended to “intimidate the public or a section of the public”, and/or to “influence the government or an international organisation”: Terrorism Act 2000, s1. The 2013 murder of Mohammed Saleem was terrorism because the accompanying campaign of mosque-bombing suggested that it was intended to “intimidate .. a section of the public”. The Finsbury Park attack was also assessed to be terrorism by CT Policing and MI5. 4 (a) Unlike the large, directed Islamist plots characteristic of the last decade, all four attacks were committed by lone actors16 or small groups, with little evidence of detailed planning or precise targeting. (b) Strong gun controls in the UK mean that bladed weapons are more commonly used than firearms in gang-related and terrorist crime.17 (c) Since a truck killed 86 innocent people in Nice (July 2016), vehicles – which featured in three of the four attacks under review – have been increasingly used as weapons.18 (d) The combination of a vehicle and bladed weapons, seen at Westminster and London Bridge, had previously been used to kill the soldier Lee Rigby (Woolwich, 2013). (e) Explosives, used in Manchester, were the most popular weapon for Islamist terrorists targeting Europe between 2014 and 2017.19 The explosive TATP has proved to be capable of manufacture (aided by on-line purchases and assembly instructions) more easily than was once assumed. Attacks thwarted 16 17 18 19 20 1.9 It is important to take the right lessons from attacks that the authorities are unable to stop. But anyone doing so needs to have in mind also the recent successes of MI5 and the police in averting terrorist attacks. A full picture of what needs to be done comes from knowing what worked, as well as what did not.20 1.10 To that end I asked MI5 and the police to brief me on the 20 Islamist-inspired terrorist plots that have been disrupted since October 2013, seven of them subsequent to the Westminster attack of 22 March 2017. They gave me that briefing in the form of a highly classified account that has been annexed to the capping document. I have had regard to the lessons learned from these incidents Having seen a number of exchanges in which terrorists revel in the apparent glamour of the “lone wolf”, I prefer the more prosaic term lone actor (or lone operator). Edged weapons were to be used in 35 of the 142 Islamist plots in Western Europe between January 2014 and mid-May 2017: Simcox, op. cit. Edged weapons caused casualties in 19 of those plots, but only three deaths and 28 injuries in total. Vehicles were to be used in just six of the 142 West European plots between January 2014 and mid-May 2017, but these plots had a 100% success rate: Simcox, op cit. Between mid-May and 22 October 2017 there were five more vehicular attacks in Europe; vehicular attacks have caused 45% of all injuries and 37% of all deaths in Islamist plots since January 2014: Robin Simcox, “Can we stop vehicular attacks?”, Lawfare blog, 22 October 2017. As this report was being finalised on 31 October 2017, a further vehicular attack took place in New York. Explosives were to be used in 40 of the 142 Islamist plots in Western Europe between January 2014 and mid-May 2017: Simcox, European Islamist Plots and Attacks since 2014 – and How the US Can Help Prevent Them, The Heritage Foundation Backgrounder, 1 August 2017. Though only six of these plots were completed, they included Paris (2015) and Brussels (2016) as well as Manchester (2017). Together, they resulted in 188 deaths and 808 injuries. For one danger of the selection bias that can result from focusing only on completed attacks, see 2.78 below. 5 which include, in particular, the importance of being able to use a wide range of complementary actions in varying combinations at different times, in conjunction with key partners both within the UK intelligence community and internationally. 1.11 It is impossible to annex a summary of each of the 20 plots to this report, in particular because of the risk of prejudicing the outcome of criminal proceedings yet to be resolved. Instead, I attach at Annex 2 a summary of the six attack plots since October 2013 that have already resulted in criminal convictions for preparing acts of terrorism and/or conspiracy to murder. The plotters had discussed or planned the use of a variety of weapons, including explosives, firearms and knives. The seriousness of those six plots is evident from the fact that a total of 10 life sentences have been imposed in relation to them. One further plot has resulted in a conviction for a lesser offence, and the remainder are the subject of pending criminal proceedings. 1.12 I also requested briefing from the police in relation to the recent frequency of XRW attack-planning. They pointed to instances of attack-planning over the 12 months to October 2017, including the construction of viable explosive devices and the acquisition of firearms, and told me that there were individuals who were assessed to have both the knowledge and the resources to carry out their desired activities. The known level of attack-planning was however very much lower from the XRW than from Islamists: by way of illustration, at the time of my detailed briefing on the XRW threat in summer 2017, there were two pending police operations into XRW attack-planning, both relatively small-scale. It was not possible to quantify the number of thwarted XRW terrorist attacks since October 2013, in part because of uncertainty as to whether a lone actor was actually planning an attack and, if so, whether it would have crossed the threshold from hate crime to terrorism. The reviews under assessment 1.13 The detailed reviews that underlie this report are concerned not with political or even policy questions, but with operational ones. Summarised in a “capping document”: (a) the post-attack reviews examine, exhaustively, the way in which intelligence was handled prior to the four attacks of March-June 2017; and (b) the Operational Improvement Review (OIR) identifies operational changes that should improve the future performance of MI5 and the police. 1.14 Highly classified, extending to some 1150 pages and containing 126 recommendations, the nine reports constitute one of the most detailed examinations ever conducted of the UK’s counter-terrorism (CT) machine and its operation. 6 1.15 They are, of course, internal rather than independent reviews. This brings certain advantages: an intimate understanding of the subject-matter, coupled with a sense of full ownership by MI5 and CT Policing of the conclusions, which – subject to resources – should ensure their complete and uncomplaining implementation. 1.16 But internal processes have potential downsides: complacency, the concealment of unpalatable facts and slowness to see the need for change. It is no doubt with such concerns in mind that as a former Independent Reviewer of Terrorism Legislation, used to conducting reviews of my own, I was asked by the Home Secretary to assure the quality of these reviews conducted by others. 1.17 This report, finalised on 2 November 2017, is the culmination of that assurance process. The reviews are more fully summarised at chapter 3 below, and the process that I undertook for assessment of them is described at chapter 4. My conclusions and some final reflections are in chapter 5. Terminology 1.18 While police terminology will be familiar to most readers of this report, the same will not necessarily be true of the vocabulary used by MI5 to describe its operations. There follows a brief definition, approved by MI5, of some of the key terms. 1.19 A lead is the term to describe intelligence or information that following initial assessment (or triage) conducted by MI5 and CT Policing, suggests a possible risk to national security. Leads are typically generated through triaging new intelligence not related to an existing investigation, though intelligence from existing investigations can also be referred for consideration as a new lead or priority investigation. 1.20 The Intelligence Handling Model (IHM) is the process for lead identification, assessment, decision-making and resolution. It has been jointly developed by MI5 and CT Policing since 2011 to manage all new CT lead intelligence and threat reporting not linked to ongoing CT investigations. It ensures that new leads benefit, where appropriate, from the combined expertise of MI5 and CT Policing. It aims to ensure that finite covert investigative resources are directed against the most credible new leads, and that leads lacking credibility are resolved in an appropriate way. 1.21 A trace is a request for a check across MI5 databases to determine whether MI5 holds adverse information, or to check that an individual is not known to MI5 already. A liaison trace is a request from a liaison partner, such as a foreign intelligence service, to determine whether MI5 holds any information relevant to their investigation(s). 7 1.22 MI5 investigations are given a priority according to the risk they carry. Priority 1 investigations are investigations into individuals or networks where there is credible and actionable intelligence of significant (P1a) or smaller scale (P1b) attack planning. P2H investigations are of high risk extremist activity linked to attack planning (e.g. a serious intent to travel overseas to undertake fighting with an extremist group, or large-scale fundraising). P2M investigations are of medium risk extremist activity not directly linked to attack planning (e.g. supply of false documents, or smaller-scale fundraising). P3 is for investigations or networks that require further action to determine whether they pose a threat, and P4 is for individuals (such as released terrorist prisoners) who have previously posed a serious threat to national security and where there is judged to be a risk of reengagement. 1.23 Tiers reflect the position and importance of SOIs within the investigation that they are assigned to. The main targets of an investigation are graded Tier 1, their key contacts are Tier 2 and other contacts of Tier 1 and Tier 2 targets, who are likely to be involved in only marginal aspects of the activities under investigation, are Tier 3. 1.24 Subjects of Interest (SOIs) are persons in respect of whom MI5 has created a Key Information Store record (or in common parlance, a file). There are around 3000 active SOIs, who are either associated with MI5 priority investigations or have come to MI5’s attention as part of a lead generated through new intelligence not part of an existing investigation. Each active SOI record is subject to quarterly case review and has an assigned lead investigator responsible for reviewing incoming intelligence and maintaining the record, including by updating it as required. 1.25 Holding codes are used to categorise the threat (if any) that each SOI is assessed to pose to national security. 1.26 Closed SOIs, as referred to in this report and in the underlying reviews, are the 20,000 or so people who have been part of MI5 Priority Investigations since 2009 but who have been given a holding code which indicates that they are no longer assessed to represent a national security threat. Other closed SOIs have been part of Lead Development investigations or have been transferred from pre-2009 legacy systems. When SOIs are closed they are categorised according to the amount of residual risk, from low to high, that they are likely to pose in the future. Rules and processes introduced since 2015 are intended to provide assurance that records for closed SOIs are being properly closed, and to mitigate the risk of new intelligence going unread as a result of attaching itself to an SOI record that is not being actively monitored. 8 1.27 21 The Annexes to this report include a list of acronyms (Annex 1) and a fuller description of MI5’s investigative processes (Annex 5), originally supplied by MI5 to the Intelligence and Security Committee of Parliament (ISC) and reproduced by the ISC as an Annex to its public report into the killing of Lee Rigby.21 That account is reproduced with some minor updates to reflect changes in practice since 2014. ISC, Report on the intelligence relating to the murder of Fusilier Lee Rigby, HC 795, November 2014, Annex A. 9 2. PRE-ATTACK INTELLIGENCE Scope of this chapter 2.1 It is not the function of this report to detail the intelligence that was available to MI5 and the police prior to the attacks. But an outline of the pre-attack intelligence picture is desirable, if my assessment of the reviews is to be understood in its proper context. 2.2 The outlines in this chapter are mostly paraphrased from the MI5 post-attack reviews, which with the help of an inspector from the Investigatory Powers Commissioner’s Office (IPCO) I have been able to verify as a true reflection of the primary source documents that they reference. To the extent that the outline is based on police pre-attack reviews (for example, as regards the immigration history of Rachid Redouane), I have not been in a position to conduct such verification. 2.3 I have not sought to satisfy public curiosity about every aspect of these operations, which would be impossible in any event for the reasons explained below. My more limited aim has been to give an idea of the quantity and quality of the intelligence that was available on each attacker at material times, thus introducing the context that needs to be understood before sense can be made of the recommendations in the various internal reviews. Transparency and its limits 22 23 2.4 There is a strong public interest in it being generally known what powers may lawfully be used by the authorities, and the extent to which such use is liable to impinge upon people’s rights and freedoms.22 The public also needs access to expert independent assessments of whether the use of intrusive powers has been lawful, necessary and proportionate.23 These are the minimum requirements for a democratic debate on the powers by which we are policed, and the manner in which those powers are used. 2.5 But it is quite a different matter to disclose how such capabilities were deployed in specific operations, particularly in a threat environment when other operations of This was a central theme of my report “A Question of Trust” (2015), subsequently reflected in the Investigatory Powers Act 2016 and the detailed Codes of Practice and information documents that accompany it. Such assessments will be provided in future by the judge-led and technically skilled IPCO. Like its predecessors, the Interception of Communications Commissioner’s Office, the Office of Surveillance Commissioners and the Intelligence Service Commissioner, IPCO will produce regular public reports. 10 a similar nature may still be in progress.24 As the ISC stated in its 2014 report following the murder of Fusilier Lee Rigby: “[T]here are some matters which we cannot include in a public report, since to do so would either be illegal or would severely damage the Agencies’ ability to protect the UK.”25 2.6 The ISC in this connection referred specifically to: a) telephone intercepts, information relating to which may not lawfully be published; b) human intelligence sources (or CHIS), whose lives may be endangered if they are known or suspected to have been operating in a particular milieu; c) operational techniques whose effectiveness would be reduced if they were generally known; and d) intelligence provided by overseas agencies, which cannot be disclosed without their permission. 24 25 2.7 Those difficulties of principle are of potential relevance to any public report on intelligence-handling, including this one. For the reasons given by the ISC, my scope to refer to the precise source of intelligence in any of the cases with which I am concerned is strictly limited. 2.8 Two other factors dictate caution in relation to the present exercise. 2.9 First, Darren Osborne (Finsbury Park) will be the subject of criminal proceedings. An extradition request to the Libyan authorities has also been made for Hashem Abedi, the brother of Salman Abedi (Manchester), with a view to his facing trial in England. Information or comment that could prejudice future trials should not be published. Accordingly, I have been able to say little about either of these individuals, or about the immediate background to the Manchester and Finsbury Park incidents. 2.10 Secondly, inquests will be held into the deaths of the attackers and their victims. The scope of those inquests (and any related inquiries) is as yet unknown. I am specifically enjoined in my letter of instruction from the Home Secretary to ensure that nothing in the public version of my conclusions could prejudice the inquests. See, by way of illustration, my Report of Bulk Powers Review (2016), in which the ambit of the relevant powers is described relatively fully but in which the practical illustrations of those powers in Appendices 58 were necessarily imprecise. ISC, Report on the intelligence relating to the murder of Fusilier Lee Rigby, HC 795, November 2014, p.2. The ISC was made a Committee of Parliament by the Justice and Security Act 2013, which increased its remit to include operational activity and the wider intelligence and security activities of Government. 11 The coroner or jury in each of those inquests will make their own determinations within their remit. Those determinations will be based on all the evidence given in the inquests. Nothing in this report is intended to affect the conduct of the inquests or their determinations. Khalid Masood (Westminster) Masood: summary 2.11 Khalid Masood, though previously known both to the police (for offences of violence prior to 2003) and to MI5 (for association with extremists, particularly between 2010 and 2012) was a closed SOI at the time of his attack.26 No intelligence was being gathered on him, and neither MI5 nor the police had any reason to anticipate the attack. Masood: personal life 2.12 Khalid Masood was born Adrian Russell Elms in 1964, and grew up in Kent. He also used the surname Ajao, and in 2005 changed his name to Khalid Masood, having converted to Islam some years earlier. 2.13 Khalid Masood had two children from a relationship in the 1990s, and two others from his second marriage, in 2007. In 2005-06 and again in 2008-09 he taught English in Saudi Arabia. He lived variously in Eastbourne, Crawley and Luton, before settling in Birmingham in 2012. 2.14 Prior to the attack in Westminster on 22 March 2017, both the police and MI5 had some limited (and largely historic) knowledge of him. This is shortly summarised below. Masood: police history 26 2.15 The majority of intelligence held on Khalid Masood was crime-related. He was convicted seven times between 1983 and 2003 for offences ranging from criminal damage to possession of an offensive weapon, threatening behavior, assault occasioning actual bodily harm, assault on police and unlawful wounding (a stabbing in the face with a flick knife). It is believed that while serving a two-year prison sentence for the latter offence, imposed in July 2000, he converted to Islam. 2.16 Significant crime-related intelligence from Sussex Police databases between 2000 and 2003 details Khalid Masood’s involvement in drug dealing, racketeering and enforcement. For an explanation of closed SOIs, see 1.26 above. 12 2.17 Khalid Masood’s final release from prison (from a six-month sentence for possession of an offensive weapon) was in 2003, when he was 38. Since that time, there is no record of him committing any criminal offence. Masood: MI5 history 27 28 29 2.18 Khalid Masood had been the subject of active investigation by MI5 between February 2010 and October 2012. At the time of the attack he was a closed SOI.27 2.19 The first trace of Khalid Masood in MI5 records dates to April 2004, when a telephone number later associated with him appeared in the contacts list of an SOI in an MI5 and police operation that investigated and subsequently disrupted a UKbased terrorist network that aimed to produce and use home-made explosives in the UK. Khalid Masood appears to have had no direct connection with that plot. Other selectors used by him,28 but not attributed to him until 2010, appeared on the periphery of various spin-off investigations between 2004 and 2009, via contact with a long-standing associate. 2.20 Khalid Masood came more fully to the attention of MI5 in 2009, on the edge of investigations into a group of individuals seeking to travel to the Federally Administered Tribal Areas of Pakistan (FATA) for al-Qaida-linked terrorist training. Khalid Masood was based in Saudi Arabia for extended periods, and it was thought that he could be identical with an individual in Saudi Arabia who was thought to be in a position to facilitate such travel. Though it later turned out that Khalid Masood was not that individual, he was placed under active investigation as an SOI for the first time on 17 February 2010, so as to enable his full identification and assess his level of involvement in extremism. 2.21 The intelligence that prompted investigation of Khalid Masood was low-level, uncorroborated reporting. Khalid Masood was not part of the principal reason that any of the investigations mentioned above was of concern to MI5, and was one of many individuals of similar profile that were investigated during that period. 2.22 On 2 March 2010, two weeks after he had become an SOI, Khalid Masood was downgraded from a holding code under which he was assessed to pose a threat to national security to a holding code under which it was assessed that he might pose a threat to national security.29 In December 2010 a review of the operation with which he had been linked recommended that Khalid Masood should be closed as an SOI. None of the intelligence relating to him suggested attack planning aspirations. He was formally closed as an SOI in October 2012. For an explanation of closed SOIs, see 1.26 above. Examples of selectors are phone numbers and email addresses. For an explanation of holding codes, see 1.25 above. 13 2.23 Between 2012 and 2016, Khalid Masood appeared intermittently as a contact of a number of SOIs, including individuals linked to the proscribed organisation ALM in Luton and Crawley. There is no intelligence to suggest that Khalid Masood was an ALM member, though in 2013 he was known to have expressed contentment that violent actions such as the World Trade Center attacks attracted people to Islam. Neither Khalid Masood’s contacts nor those comments were considered to reach the threshold for re-opening an investigation into him. Masood: post-attack intelligence 2.24 The post-incident investigation commenced immediately after the Westminster attack and shed light on Khalid Masood’s preparation for his attack. 2.25 He researched violent attacks, knives, Daesh and vehicle types online as early as April 2016. 2.26 Between December 2016 and March 2017 he informed members of his family that he was considering working overseas, but his job and visa applications were unsuccessful. He sold his car and made efforts to say goodbye to members of his family. 2.27 On 9 March 2017, Khalid Masood purchased two Sabatier carving knives from Tesco in Birmingham, and on the same day sent himself an email with the subject line “Retaliation”. On 15 March he was in possession of a document entitled “Jihad in the Quran and Sunnah”, with his photograph on the front page and multiple extracts from the Quran that could be claimed to be supportive of jihad and martyrdom. Having made a down-payment on 8 March, he collected on 16 March the Hyundai Tucson that was used in the attack. 2.28 On 19 March, Khalid Masood conducted reconnaissance of Westminster Bridge in person and online, and browsed YouTube for videos relating to terrorism (including suicide attacks). Further similar browsing was noted over a variety of dates. Subsequent searches of various media devices belonging to him concluded that he appeared to advocate a conservative, Saudi-influenced, Salafist interpretation of Islam, but noted that his relatively small digital collection did not contain much of the standard jihadi content that is normally found in investigations involving Islamist-inspired terrorists. 2.29 A few minutes before the attack on 22 March, Khalid Masood shared his Jihad document with numerous WhatsApp contacts; shortly afterwards it was sent via iMessage and SMS to additional contacts. 14 Salman Abedi (Manchester) Abedi: summary 2.30 Like Khalid Masood (but 30 years younger), Salman Abedi was a closed SOI at the time of his attack, and so not under active investigation. MI5 nonetheless came by intelligence in the months before the attack which, had its true significance been properly understood, would have caused an investigation into him to be opened. It is unknowable whether such an investigation would have allowed Abedi’s plans to be pre-empted and thwarted: MI5 assesses that it would not. 2.31 Salman Abedi was also identified by a separate data-washing exercise (see 2.38 below) as falling within the small number of closed SOIs who most merited further consideration. Unfortunately, the timing of that exercise was such that the meeting scheduled to consider the results of this process had not been held as of the date of the attack. Abedi: personal life 2.32 Salman Abedi was born in Manchester in 1994, to parents who had been granted asylum after fleeing the Gaddafi regime in Libya. He was the second of six children, the third being his brother Hashem who is currently in detention in Libya and the subject of an extradition request. 2.33 Given the possibility of a trial, nothing more is said here about Hashem or about the pre-attack phase (which involved the purchase of ingredients and the manufacture of an explosive device), save to note that Salman Abedi was in Libya between 15 April and 18 May 2017, four days before the attack. Abedi: police history 2.34 Salman Abedi’s criminal record is limited to reprimands for theft and receiving stolen goods in 2012, and an assault on a female while at college which was dealt with by restorative justice. Abedi: MI5 history 2.35 Salman Abedi was first actively investigated in January 2014, when it was thought that he might have been an individual who had been seen acting suspiciously with an SOI. Although he knew the SOI in question, he turned out not to have been the individual seen with him, and his record was closed in July 2014. He was classed as a closed SOI of low residual risk, given his limited engagement with persons of national security concern. 15 2.36 Salman Abedi was again opened as an SOI in October 2015, on the basis of his supposed contact with a Daesh figure in Libya, but he was closed as an SOI on the same day when it transpired that any contact was not direct. 2.37 Although he remained a closed SOI until the day of the attack, Salman Abedi continued to be referenced from time to time in intelligence gathered for other purposes. On two separate occasions in the months prior to the attack, intelligence was received by MI5 whose significance was not fully appreciated at the time. It was assessed at the time to relate not to terrorism but to possible nonnefarious activity or to criminality on the part of Salman Abedi. In retrospect, the intelligence can be seen to have been highly relevant to the planned attack. 2.38 Another tool promised well, but did not produce results in time. A process devised by MI5 to identify activity of renewed intelligence interest conducted by closed SOIs, using targeted data exploitation and other automated techniques, identified Salman Abedi as one of a small number of individuals, out of a total of more than 20,000 closed SOIs, who merited further examination. A meeting (arranged before the attack) was due to take place on 31 May 2017: Salman Abedi’s case would have been considered, together with the others identified. The attack intervened on 22 May. Khuram Butt (London Bridge) Butt: summary 2.39 Khuram Butt, uniquely among the protagonists in the attacks under review, was a live SOI, under active investigation at the time of his attack. He was the principal subject of an MI5 investigation which I will refer to as Operation HAWTHORN,30 opened in mid-2015 following information suggesting that he aspired to conduct an attack in the UK. Coverage of various kinds was put in place over a period of almost two years. Though it continued to varying degrees until the day of the attack, it did not reveal the plans of Khuram Butt and his two co-conspirators. Butt: personal life 30 2.40 Khuram Butt was born in Pakistan in 1990. His family moved to England in 1998, claiming asylum based on political oppression: they were given indefinite leave to remain in 2004, and Khuram Butt was given British citizenship in 2005. 2.41 He was schooled in Forest Gate, East London, and attended a local sixth form college. Between 2012 and 2015 he worked as an office manager with a subsidiary of KFC. In 2013 he married the sister of a friend. The couple had a Not being authorised to use the real name of this operation, I follow the course taken by the ISC, in its report into the murder of Lee Rigby, of using instead the name of a tree. 16 son born in October 2014 and a daughter born in May 2017, less than a month before the London Bridge attack. 2.42 Khuram Butt made a pilgrimage to Mecca in February 2015. He expressed frequent aspirations to travel from late 2015, including to Syria, but never again left England. Butt: police history 2.43 Khuram Butt had no criminal convictions. He received cautions for offences in 2008 and 2010. 2.44 In January 2016, he was identified posing with a Daesh flag in the Channel 4 television documentary, “The Jihadis Next Door”. The police reviewed the documentary and deemed that no criminal offences had been committed, a judgement subsequently endorsed by the Crown Prosecution Service (CPS). 2.45 While under investigation by MI5, Khuram Butt was arrested for fraud in October 2016 and granted bail. He had not yet been told by 3 June 2017, the date of the attack, that on 1 June the decision had been taken not to prosecute him. 2.46 Khuram Butt was identified as the suspect for a common assault against a member of the Quilliam organisation in July 2016. The victim was unwilling to press charges, and delays resulted in the statutory time limit for summary offences being exceeded. No judicial disposal was therefore possible. Butt: MI5 history 31 32 2.47 From mid-2015 until the date of the attack, Khuram Butt was the principal subject of Operation HAWTHORN, an MI5 investigation which was opened following information suggesting that he aspired to conduct an attack in the UK. As part of the prioritisation system used for ongoing operations, HAWTHORN was graded P2H, signifying high risk extremist activity linked to attack planning.31 A significant amount of coverage was put in place following the initial reporting. In September 2015, a potential lone actor triage assessment concluded that Butt represented a MEDIUM risk due to his strong intent but weak capability.32 2.48 From late 2015 to early 2016 there was no further indication of attack-planning, and Khuram Butt appeared to be disengaging from former associates in ALM. His focus seemed to be moving towards overseas travel, including potentially to Syria to fight with Daesh or to another Arabic-speaking country to learn the language. For an explanation of the prioritisation system see 1.22 above. MI5 and CT Policing assess SOIs against certain criteria to establish whether the SOI poses a lone actor threat. A risk assessment (on a scale of no to high, plus unresolved which indicates that more intelligence is required) is then agreed following discussion with MI5’s behavioural science team. 17 MI5 took steps to ensure that it would be aware of an intent to travel. HAWTHORN was suspended from February to April 2016 because of resourcing constraints in the wake of the Bataclan attack in Paris.33 Spring 2016 saw further aspirations to travel (to other countries in the Middle East and/or Africa) and to raise money for travel, but no longer any indication that travel would be for extremist purposes. MI5 took the decision not to prevent any planned overseas travel. 33 34 35 2.49 In the second half of 2016, Khuram Butt re-engaged with ALM and was believed to be involved in bank fraud. HAWTHORN was downgraded from P2H to P2M in September.34 After his arrest for fraud in October (2.45 above), Butt began a period of withdrawal from ALM and increased his operational security. 2.50 Early in 2017, Khuram Butt worked at the Ummah Fitness Centre, developed links with extremist associates (including his co-conspirator Rachid Redouane, as transpired after the attack) and was identified as teaching a Qur’an class to young people (involving in this work his co-conspirator Youssef Zaghba, as also transpired after the attack). Concerned that he would use that opportunity to radicalise, MI5 and police worked together to try to identify and disrupt this activity. 2.51 On 21 March 2017, prior to the Westminster attack on the following day, investigation of Khuram Butt was suspended. Investigation of the other SOIs investigated under the operation had been suspended the previous week, due to resourcing constraints brought on by a large number of P1 investigations. Some intelligence however continued to be gathered, and analysis performed. In April 2017, when Operation HAWTHORN was still suspended, MI5 downgraded Khuram Butt’s holding code from one that indicated he was likely to pose a threat to national security to one that indicated he might pose such a threat.35 MI5 noted his continued extremist rhetoric but also uncertainty about whether he posed a threat to national security. 2.52 On 5 May 2017 Operation HAWTHORN was re-opened, with a view to considering whether the threat needed continued investigation or whether it could be closed and resources deployed elsewhere. Though HAWTHORN was likely to close unless significant developments were identified, consideration was given to what would happen to Khuram Butt thereafter. A number of options were identified In order to focus investigative resources on the highest risks, it is sometimes necessary to suspend temporarily live investigations that MI5 assesses to carry lower risk. Suspending a live investigation is a formal decision with established processes and review mechanisms. A suspended investigation is not subject to pro-active investigative steps, and there is no regular review of incoming intelligence associated with it. Accordingly, a decision to suspend an investigation for any length of time is taken in consultation with MI5 senior managers. The grading of priority investigations is explained at 1.22 above, and in more detail in Annex 5. Holding codes are briefly explained at 1.25 above. 18 including continued active investigation as part of another operation and monitoring as a closed SOI of medium residual risk. 2.53 A second lone actor triage assessment (see 2.47 above) was completed on 15 May 2017. It concluded that the risk posed by Khuram Butt had moved from MEDIUM to UNRESOLVED and that further investigative options should be progressed in order better to understand the threat that he posed. 2.54 Material relating to Khuram Butt received in the two weeks prior to the attack on 3 June added little to the intelligence picture and did not identify the activity that led up to the attack. As of 3 June, the date of the attack, HAWTHORN remained a live investigation. Butt: post-attack intelligence 2.55 Post-incident investigation has revealed periods of co-location between the three conspirators, most notably from December 2016. It is assessed that the relationships between Redouane and Butt and (tentatively) Zaghba and Butt may have started at the Ummah Fitness Centre. A meeting there on 7 March, attended by both Butt and Redouane, appears significant and may have related to an attempt by Butt to acquire a firearm. 2.56 Khuram Butt displayed strong operational security and much remains unknown, even today, about the mindset of the three conspirators and the planning of the attack. 2.57 CCTV footage from Khuram Butt’s home address on the evening of the attack showed footage of Butt getting into a white van, hired earlier that day, with a large red holdall. Two males accompanied him. One of them carried chairs, perhaps to support a cover story that the van was required for moving furniture. Rachid Redouane (London Bridge) 2.58 The following account, which I have not independently verified against the source documents, is paraphrased from the police review. 2.59 Rachid Redouane first came to the notice of the UK authorities in 2009, when he sought asylum under the false identity of a Libyan national. Asylum was refused and appeal rights exhausted. He last reported to immigration officials in April 2011, after which he was considered an absconder. 2.60 He was stopped and arrested in June 2012, again under a Libyan name, by Police Scotland as he attempted to travel to Northern Ireland by boat. At the time removals to Libya had been suspended on humanitarian grounds, so he was released from detention in Larne with conditions to reside in Dagenham and report 19 to immigration authorities. His details were added to the Police National Computer in line with normal absconder procedures, but he was never tracked down. 2.61 Rachid Redouane had no other UK police record. Post-incident checks have revealed that he was known to Moroccan police, though not linked to terrorism. 2.62 Between 2013 and 2015 Redouane was in Morocco. In November 2014 he was issued with an Irish visa in Casablanca, using a passport which had been issued in 2013 by the Moroccan Embassy in London in the name of Rachid Redouane. In 2015 he travelled to Dublin, moved in with his Irish wife, and applied successfully for an EEA Family Permit and EEA residence card, sponsored by his wife. 2.63 At the time of the attack Redouane was living legally in the UK under his Moroccan identity, though the issue of his failed asylum claim under a Libyan identity had not been resolved. 2.64 Rachid Redouane was not investigated by MI5 or the police prior to the London Bridge attack. Before the attack, MI5 had received a number of strands of intelligence regarding a Moroccan male named “Rashid” who was assessed by MI5 to be a peripheral and social associate of Khuram Butt. Following the attack, analysis identified “Rashid” as Rachid Redouane. Youssef Zaghba (London Bridge) 36 2.65 Youssef Zaghba was a dual national of Morocco and Italy. Born and brought up in Morocco, his parents had divorced and his mother returned to Italy. 2.66 DWP records show that Zaghba had been working legally in the UK since 30 June 2015. He last entered the UK on 12 January 2017. He is believed to have been single with no dependents. He had no UK, Italian or Moroccan criminal record, and had only come to the attention of police in the UK as a witness to an assault in 2016. 2.67 Post-incident reporting shows that during a stop at Bologna Airport in March 2016, Youssef Zaghba had said that he was travelling to Turkey as a “terrorist”, but quickly changed that to “tourist”. Further investigation in Italy revealed that he had expressed an interest in travelling to Syria to join Daesh and practice the “Real Islam”. 2.68 On 23 March 2016, the Italian authorities placed Youssef Zaghba on the SIS II warning list,36 thereby potentially bringing him to the attention of the UK authorities SIS II is the second-generation Schengen Information System, an EU-wide IT system which helps facilitate European cooperation in law enforcement, border control and immigration. 20 at the border, but under a marker which identified him as subject to checks for serious crime. A different marker, which the Italian authorities did not use, would have automatically identified him as a national security risk. The marker was deleted by the Italian authorities on 23 January 2017: in the meantime, UK Border Force staff had noted him as passing through British ports on three occasions. 2.69 In June 2016, MI5 received through international channels an enquiry from the Italian authorities in relation to the incident at Bologna Airport and the subsequent intelligence about Youssef Zaghba’s desire to go to Syria. The Italian authorities requested traces on Zaghba (who they said spent parts of the year working in a London restaurant), and any contacts he had in the UK with individuals linked to Islamist extremism and/or with Italy. 2.70 MI5 has no record of responding to this enquiry, noting by way of possible explanation that it arrived in the incorrect mailbox in MI5. It was not chased up by the Italian authorities. The story is not a happy one: but as MI5 points out, even if the request had been actioned, it would have resulted in a nil return. 2.71 Youssef Zaghba was not investigated by MI5 or the police prior to the attack. Postattack analysis identified him as a user of the Ummah Fitness Centre, and as a man whom Khuram Butt had introduced in February 2017 to the class he was teaching. Darren Osborne (Finsbury Park) 37 2.72 Darren Osborne is a British citizen, born in Singapore in 1969 where his father served in the Royal Air Force. He lived in Nottinghamshire and Lincolnshire as a child, before moving to Weston-Super-Mare. In 2006 he moved to Cardiff, and has lived at his current home since 2009. 2.73 Because of the pending trial, I can say nothing on the subject of whether, and if so how, Darren Osborne was known to the police prior to the attack. Suffice it to say that police had no intelligence to suggest that he was going to commit the attack, and that MI5 – which does not lead investigations in relation to so-called “domestic extremism”37 – held no intelligence on him. 2.74 Nor can I comment on the period immediately prior to the attack, or on the extent to which the movements and motivations of Darren Osborne during that period may have been illuminated by post-attack investigations. On “domestic extremism”, see further 3.46 and 5.4(c) below. 21 General observations on the intelligence picture 2.75 The attackers had varying degrees of visibility to MI5: (a) Khuram Butt (London Bridge) had been the subject of an MI5 priority investigation for almost two years. (b) Khalid Masood (Westminster) and Salman Abedi (Manchester) were closed SOIs: Masood had previously been an SOI as a peripheral figure in a wide investigation, and Abedi had previously been an SOI for two short periods while the nature of his connections with others was cleared up. Prior to their attacks: (i) the re-engagement of Masood, who appears to have acted alone, was not picked up in any way, whereas (ii) further information about Abedi was received by MI5 which, had it been assessed differently, would most likely have triggered the opening of an investigation.38 (c) Rachid Redouane and Youssef Zaghba (London Bridge) were never investigated by MI5, but were later identified as contacts of Khuram Butt. (d) Darren Osborne appears to have acted alone and was unknown to MI5. 38 39 2.76 Closed SOIs are far more numerous than active SOIs, and as can be seen from the Westminster and Manchester attacks, some closed SOIs are well capable of perpetrating deadly attacks. Their dangers are exacerbated by the fact that closed SOIs are by definition not under active investigation (though as demonstrated in both these cases, there may be means – albeit not sufficient on this occasion – of compensating for this). 2.77 How best to deal with the risk from persons not under active investigation has been a long-standing challenge for MI5, in respect of which a number of solutions have been tried in the past.39 Many of the recommendations in the OIR, summarised in chapter 3 below, are directed to improving coverage of such persons. 2.78 Concentration on completed attacks, a number of which have involved closed SOIs, might suggest that closed SOIs are more dangerous (as well as more The process referred to at 2.38 above could also have resulted in further action, had it been completed in time. The ISC has commented on the various projects operated between 2007 and 2014 by MI5 and the police to manage what they described as “’low-level’ SoIs – individuals who are not necessarily deemed enough of a security risk to appear in a Priority 1-4 investigation, but in whom MI5 and the police still have an interest”: ISC, Report on the intelligence relating to the murder of Lee Rigby HC 795, November 2014 paras 45-55. 22 numerous) than live SOIs. This could throw into significant doubt MI5’s allocation of resource to the two groups. But it is important not to succumb to selection bias. The perpetrators of the overwhelming majority of the 20 Islamist plots foiled since mid-2013 (six of which are listed at Annex 2 below) were SOIs under active investigation: indeed it was generally those investigations that enabled their plans to be identified and thwarted. Their inclusion in the sample alters the picture significantly. 23 3. THE INTERNAL REVIEWS Summary of the reviews 3.1 Eight internal reviews (seven Post-Attack Reviews and the OIR) were conducted by MI5 and the police, summarised in a capping document which brought together their conclusions and recommendations. Collectively, these detailed and highly classified documents provide a comprehensive account of the handling of intelligence prior to each attack, and of the operational improvements that MI5 and the police have identified as desirable in the light of that experience and more generally. 3.2 The seven post-attack reports and OIR were produced in final form on 19 October 2017 and made available in the following week, under cover of the capping document, to the Prime Minister, the Home Secretary and the Chair of the Joint Intelligence Committee (JIC).40 They have also been supplied to the ISC. Together, the documents extend to some 650 pages of text, with a further 500 pages of Annexes and references. 3.3 Very considerable resource went into producing these reports, at an extremely busy time for both organisations: (a) MI5 devoted a total of 35 full-time equivalent staff to the reviews for a period of 16 weeks between June and October. (b) CT Policing devoted 12 full-time equivalent personnel to the reviews over the same period, including an Assistant Chief Constable and two Detective Superintendents. A further 11 full-time equivalent personnel worked on the reviews for periods of two weeks, and another eight had ad hoc input. (c) My own assurance function occupied 37 days of my time between late June and 2 November, together with a total of nine days from those who assisted me (4.9 below). While the value of the reviews will no doubt repay the efforts put into them, it may be useful to be aware of the extent to which they occupied resources and took personnel away from current activities. 40 The JIC, whose membership comprises senior officials from a variety of government departments, assesses national security and international developments, drawing on secret intelligence, diplomatic reporting and open source material. These assessments are provided to Ministers and departments to inform operational planning and policy action. 24 Post-Attack Reviews: specific lessons Scope of the reviews 3.4 In relation to the first three attacks, MI5 and the police each produced their own reports. Because of the nature of the investigations in question, the majority of the material relates to MI5 rather than CT Policing. In relation to the Finsbury Park attack, there is only a CT Policing report because Darren Osborne was unknown to MI5. 3.5 The terms of reference of the post-attack reviews required them to: (a) identify what was known about the attackers and alleged co-conspirators prior to each attack; (b) review assessments, actions and decisions made prior to each attack in relation to intelligence held on the attackers and co-conspirators; (c) identify and review contextual information that may have had a bearing on actions and decisions made; and (d) identify learning points arising out of each case. The focus of the post-attack reviews was therefore uniquely on the period prior to each attack: none of them is concerned with the progress of the attacks themselves or with the post-incident response. 3.6 Having summarised the facts relative to each attack, the MI5 and police reports set out in exhaustive detail the intelligence that was available prior to the attacks, the actions that were taken in response to that intelligence, and (where relevant) any light shed by post-incident intelligence. Most of them conclude by setting out learning points and recommendations.41 Conduct of the reviews 41 3.7 MI5’s post-attack reviews were conducted by a team from MI5, none of whose members had prior involvement in the matters under review. CT Policing staffed its reviews on the same basis. In both cases, the teams were led by senior officers. They drew on wider professional expertise across both institutions, with input from MI6 and GCHQ, to test and validate their conclusions. 3.8 MI5’s post-attack reviews drew on material identified through a comprehensive search of MI5’s electronic documentation system, via an established process. The The exceptions are the police reports into Westminster, Manchester and London Bridge, whose learning points are incorporated directly into the capping document. 25 primary use of this process is to find material that is potentially disclosable in legal proceedings. It involved many levels of complex search, agreed with the relevant investigative teams, against a large number of specific search terms and identifiers which it was judged could feature in relevant documentation. All documents recovered were then sifted manually by the review teams, under management supervision, to assess their relevance. Their number was substantial, particularly in the case of London Bridge, since Khuram Butt had been subject to active investigation over a significant period of time. GCHQ and MI6 conducted their own searches in accordance with an equivalent process. 3.9 CT Policing’s post-attack reviews were led by the National Counter-Terrorism Policing Headquarters (NCTPHQ). There is no single police intelligence database in existence, but rather a collection of regionally or locally held standalone intelligence systems, some of which can be accessed nationally. The police describe their searching as painstaking, against agreed search parameters, seeking relevant or potentially relevant entries. Where intelligence or other records were found, they were viewed and assessed by members of the team and decisions were made by senior officers on inclusion in the final report. 3.10 An important element of the review process was the series of “pause point” meetings at MI5, each up to a day in length and devoted to a single attack, that critically assessed the decision-making from a wide range of perspectives. I attended these meetings, some of which were specifically scheduled in order to enable me to do so. Even more intensive discussions were held in relation to a few individual topics, again in my presence. 3.11 Towards the end of the process, a senior group of practitioners from across MI5, the other Agencies and CT Policing convened two further sessions to discuss the strategic themes that emerged from the post-attack reviews, and how these were reflected in the OIR. I attended both sessions, in one case with two senior advisers (4.9(a) and (c) below) who were able to give me their reactions as informed outsiders. Westminster attack: MI5 conclusions and recommendations 3.12 The review team concluded that actions taken in relation to the threat posed by Khalid Masood, including the decision to close him as an SOI, were sound on the basis of the information available at the time. It identified, however, a number of learning points and areas where MI5’s policies and processes should be amended. 3.13 Those learning points, with accompanying recommendations, concerned the categorisation of SOIs, record-keeping and the recording of decision-making, the assessment of intelligence on existing SOIs and the handling of intelligence regarding violent behaviour. 26 Manchester attack: MI5 conclusions and recommendations 3.14 The review team concluded that the investigative actions taken in relation to Salman Abedi in 2014, and the subsequent decision to close him as an SOI, were sound on the basis of the information available at the time. It identified several further examples of good practice, including the “tripwire” coverage that remained after he had ceased to be an SOI (2.37 above) and the use of the tool described at 2.38 above. 3.15 Detailed consideration was given to the way in which MI5 in early 2017 handled the intelligence whose true significance was not appreciated at the time (2.37 above). On this, the review team concluded in summary that: (a) the decision not to re-open an investigation was “finely-balanced”, and “understandable” in the circumstances; (b) there is a high degree of inherent uncertainty in speculating as to what might or might not have been discovered if an investigation had been opened on the basis of the new intelligence; but that (c) on the clear balance of professional opinion, a successful pre-emption of the gathering plot would have been unlikely. 3.16 It was also noted that despite his status as a closed SOI, an opportunity was missed by MI5 to place Salman Abedi on ports action following his travel to Libya in April 2017. This would have triggered an alert when he returned shortly before the attack, which could have enabled him to be questioned and searched at the airport by CT Policing under Schedule 7 to the Terrorism Act 2000. 3.17 A number of learning points and recommendations were identified. These concerned the handling of closed SOIs, triaging intelligence and the leads processing system, handling potential high-risk intelligence with an indeterminate terrorist threat, key investigative judgements, the use of travel notification/monitoring tools, record-keeping (including considerations of automation) and the processes surrounding certain types of information-gathering. London Bridge attack: MI5 conclusions and recommendations 3.18 The review team concluded that the investigation into Khuram Butt was run well and that the decisions taken in the investigation were sound in the light of what was known at the time. A number of areas of good practice were highlighted, including the rapid allocation of resources after the initial threat reporting in mid2015, and the consideration given to disruptive options to prevent travel. The suspensions of Operation HAWTHORN (2.48 and 2.51 above) were judged to be 27 appropriate bearing in mind the limited resource available to detect and disrupt high priority threats. 3.19 Though a number of learning points were once again identified, the review team concluded that none of the points identified resulted in missed opportunities to detect and prevent the attack. The learning points, and associated recommendations, concerned thematic investigation management, the potential lone actor triage process (2.47 and 2.53 above), the suspension and re-opening of live investigations, disruption, plans to close HAWTHORN (2.52-2.53 above), responses to liaison traces (2.69-2.70 above) and a number of points arising out of the post-incident investigation. All attacks: police recommendations 3.20 The police grouped together their learning points and recommendations from the four attacks, combining them with broader reflections about the way CT Policing and MI5 work together. They are set out, together with all the MI5 recommendations, in an Annex to the capping document. 3.21 Those learning points and recommendations concern the anti-terrorist hotline, the deployment of CHIS, community engagement, operational information management and data standards, the application of the IHM,42 investigative strategy, the use of IT to support investigative processes, the monitoring and flagging of closed SOIs, the use of powers relating to travel, mechanisms for reporting the purchase of chemicals, lone actor case work, prisons and the need for parity in the threat assessment and investigation of different types of terrorism. 3.22 The police’s Finsbury Park post-attack review additionally reflects in detail on whether the police could conceivably have done anything to prevent the attack, as well as touching on procedures that might in future be put in place when (for example) applying for an HGV licence or hiring a vehicle. Post-attack reviews: strategic themes 3.23 The review teams took the opportunity to stand back from the individual attacks and identify themes of common interest, which they set out in the capping document under the following heads: (a) the shift in the threat, and its consequences for the volume of leads and the prioritisation of resources; (b) the challenges of detection, among them the increasingly swift development of threats, operational security on the part of targets and their 42 For an explanation of the IHM, see 1.20 above. 28 use of encrypted applications (as was noted in the Westminster, Manchester and London Bridge post-attack investigations); (c) extremist material online, including instructional videos and propaganda; (d) explosive precursors, and the need to identify purchases of potential national security concern; (e) the feasibility of identifying vehicle hire for the purposes of terrorism; (f) processes, including various improvements to the IHM that is used by MI5 and police to prioritise investigative activity, and the review of guidance; (g) use of disruptive powers including criminal prosecution and the use of executive orders; (h) use of powers relating to travel, including the power to question at ports and airports; (i) ALM, the proscribed organisation in whose circles Khuram Butt (London Bridge) and to a lesser extent Khalid Masood (Westminster) moved at various times. (j) the overall investigative approach, with particular reference to proportionality, the adequacy of guidance, and the issue of whether officers had “self-censored” on the basis of concerns about proportionality, litigation risk or capacity; (k) the extent to which investigatory powers proved useful, including both human sources, physical and audio surveillance, equipment interference, interception and the two “bulk powers” that were used in these investigations: bulk personal data and bulk communications data;43 and (l) the exploitation of intelligence opportunities which arise when SOIs are in prison. They also looked at the management of so-called domestic extremism casework. 3.24 43 Many of these strategic themes fed into the OIR, discussed below. They continue also to feed into the current revision of the CONTEST counter-terrorism strategy, and into wider thinking on extremism and integration. The nature and utility of those powers are described in my Report of Bulk Powers Review, August 2016. The two other bulk powers reserved to security and intelligence agencies under the Investigatory Powers Act 2016, bulk interception and bulk equipment interference, were not relevant to these cases: they may be applied for only when the main purpose is to obtain overseas-related communications, information or certain types of data. 29 Operational Improvement Review Origins and purpose of OIR 3.25 The Operational Improvement Review (OIR), commissioned by the Commissioner of the Metropolitan Police and the Director General of MI5 after the Westminster and Manchester attacks, was already under way at the time of the London Bridge and Finsbury Park attacks. 3.26 As its name suggests, the OIR was to be operationally focused. Its purpose was to identify and take forward improvements to processes and capabilities for managing terrorism investigations. Scope of OIR 3.27 According to its final terms of reference (Annex 3), which were agreed by the OIR Steering Board (OIRSB: see 3.33 below), the aim of the OIR was to: (a) further improve operational effectiveness, building on the strong counter terrorism machinery that has been built in the UK over more than a decade; and (b) provide MI5 and the Metropolitan Police’s input on this issue into any wider reviews of the UK’s counter terrorist and counter extremism systems. 3.28 The scope of the OIR evolved as time went by, so as to ensure that the correct lessons were learned from each of the four attacks under review. Given the short timescales, it was acknowledged that some of the recommendations could only be for further work to be taken forward following completion of the OIR. 3.29 Seven specific matters were identified in the terms of reference for consideration: (a) whether there are any further improvements that can be made in how leads, prioritisation and triage processes operate; (b) the process by which individuals are categorised as closed subjects of interest, and how cases are then reviewed and escalated where indicators of potential re-engagement in terrorist activity are identified; (c) what data sources, tools and approaches, both tactical and strategic, can best support this work; (d) which partners can contribute, and how we can further strengthen joint working in managing this risk; 30 (e) how data is shared and links are made with other organisations / interventions (such as PREVENT or community policing), where appropriate; (f) any policy, legal or ethical questions associated with this issue that require consideration; and (g) the overall level of assurance it is possible to provide in this area, and the resource and prioritisation challenges associated with managing this risk. 3.30 Its terms of reference emphasise the limited nature of the OIR, which was “not intended to examine the entire investigative process or to replicate broader issues being considered in reviews of CONTEST or CT legislation”. The OIR was specifically directed not to: (a) make recommendations on thresholds for and allocation of intrusive resources, or the development of capability and processes in Priority investigations; (b) examine the suspension of Priority investigations where resources (including investigators) are required for more urgent investigations; (c) examine the balance of priority investigative casework across Thames House and the Regions; or (d) review the processes for managing priority investigative casework. 3.31 The OIR was deliberately focused on the areas where there was likely to be greatest scope for improvement of processes, in particular the triage of incoming leads, the prioritisation of investigations and the management of residual risk posed by individuals no longer under investigation. With the benefit of hindsight, that focus was a little narrow. It was always likely that the post-attack reviews would throw up lessons of a general nature falling outside the rather specific OIR terms of reference; and so indeed it proved. 3.32 With my encouragement, MI5 and the police sought not to downgrade or ignore such lessons but to accommodate them in the OIR vehicle. Thus, by agreement: (a) The OIRSB agreed that the handling of so-called domestic extremism would be considered by the OIR, given the Finsbury Park attack. (b) A number of further recommendations identified by the police in the postattack reviews, relating for example to prisons and borders, were included 31 in an Addendum to the OIR: while not considered by the OIRSB, they were discussed with and agreed by affected agencies and departments. (c) The principal strategic themes encountered in the post-attack reviews were separately identified in the capping document, as detailed at 3.23 above. Conduct of OIR 3.33 The OIR review team was jointly led by senior members of MI5 and CT Policing, and included representatives also from the Home Office and GCHQ. The review team reported to the OIRSB consisting of senior representatives of the Security and Intelligence Agencies, Police, NCA and relevant Whitehall Departments, with input from MI5’s Non-Executive Directors. The timescale was short, as required by the terms of reference. 3.34 The OIR team liaised with the post-attack reviews to ensure that their lessons were reflected in the OIR where relevant. This was an iterative process, given that the time-consuming investigation of London Bridge in particular meant that the lessons learned from it were not fully apparent until a relatively late stage in the review process. Recommendations of OIR 3.35 The OIR is an operational document. For that reason alone, much of the detail of the recommendations is necessarily sensitive and cannot be discussed here save in broad outline. 3.36 Many of its conclusions and recommendations relate to the refinement of processes in the engine room of MI5 and police CT investigations. Some were suggested by the post-attack reviews; others draw on existing workstreams for continuous improvement and for the stepping up of MI5 capabilities in line with the budget increase that followed the 2015 Strategic Defence and Security Review (SDSR): see 5.25 below. 3.37 The recommendations of the OIR comprise, broadly, three step changes and a variety of smaller but still significant improvements to current practices. Step Change (1): Exploiting Data 3.38 The first step change relates to improvements in the ability of MI5 and police to exploit data to detect activity of concern, particularly on the part of closed SOIs but in relation also to active SOIs and previously unknown individuals. 3.39 Two specific strands to this work are proposed: 32 a) a better strategy for acquiring, analysing and sharing data across intelligence and policing, for example through wider use of bulk personal datasets and by enhancement of tools such as that referred to at 2.38 above; and b) increasing cooperation with the private sector, for example to improve the detectability and even the preventability of purchases of potential explosives precursors by would-be terrorists, as undertaken before the Manchester and Parson’s Green attacks. Step Change (2): Multi-Agency Engagement 3.40 The OIR proposes going beyond existing relationships and processes to enable the widest range of partners to play their part in managing the risk of renewed engagement in terrorism-related activity by closed SOIs. 3.41 Key to this development is a new commitment by MI5 to allow knowledge derived from intelligence to be shared more widely beyond intelligence circles. This should enable, for example, neighbourhood policing and other agencies to make judgements with a better knowledge of the national security risk, and to implement appropriate local action. Improved flow of information in the other direction also has the potential to contribute to better decision-making at the centre in relation to the risk from closed SOIs in particular, for example by refining the tool described at 2.38 above. 3.42 I am unable to give further detail, not least because these proposals have still to be linked into wider work being considered under the Government’s review of CONTEST, and so have not yet taken final shape. Step Change (3): Domestic Extremism 3.43 The third step change proposed in the OIR is an increased role for MI5 and for JTAC in so-called domestic extremism work, including in particular XRW terrorism. The aim is to ensure the equivalence of processes in analysing and dealing with all kinds of terrorism, irrespective of the ideology that inspires them. 3.44 The assessment of different types of terrorist threat has been conducted in the past by different bodies. The OIR proposes, with the agreement of CT Policing and of MI5, that JTAC should in future be responsible for producing national Threat Assessments of terrorist threats arising from so-called domestic extremism work. JTAC’s involvement in its assessment will ensure a common language, methodology and approach. 33 3.45 In addition, it was determined in the OIR that there should be a formal dialogue and structures to enable MI5’s involvement in the relatively small proportion of domestic extremism work that crosses the statutory threshold for terrorism. 3.46 Finally under this head, it was recommended that consideration be given to putting an end to what some have considered the outdated and confusing terms “domestic extremism” and “international counter-terrorism”.44 The law treats different kinds of terrorism equally. So, the argument goes, should the terminology applied to them by intelligence agencies and police. Other recommendations 3.47 Alongside those step changes are a substantial number of more detailed recommendations for improvement, some of them technical in nature and many of them drawing on strands of work that pre-dated the attacks under review. These include: (a) continuing improvements to the handling of intelligence and alignment of processes between MI5 and CT Policing, with a particular focus on ensuring consistency across the regions and in the way that new intelligence is assessed for the purposes of leads, triage and prioritisation; (b) detailed recommendations relating to improvements to the systems, guidance and practices in place for information management; (c) the continued development of a joint MI5/CT Policing lone actor strategy, in order to ensure a common understanding of the risk framework and a joined up approach to the mitigation of threats; (d) the reinforcement of processes for PREVENT referrals, to ensure that all suitable candidates are referred and to ensure consistency in processing of referrals, recognising the statutory duties of other organisations in relation to safeguarding and data protection; (e) a continued focus on the challenges posed by encryption; and 44 “Domestic extremism” has been used as a catch-all category for a range of activities ranging from criminal acts associated with lawful protest through to acts not perpetrated by Islamists which meet the statutory definition of terrorism. The concept of extremism is notoriously hard to pin down (for a judicial attempt to do so, see Shakeel Begg v British Broadcasting Corporation, [2016] EWHC 2688 (QB), Haddon-Cave J). Police told me of confusion caused by the existence of three working definitions of “domestic extremism” (from the Association of Chief Police Officers (ACPO) (2006-2011), the PREVENT strategy (2011) and the National Domestic Extremism Unit (NDEU) (2013)), and added that a call in 2012 by HM Inspectorate of Constabulary for an agreed definition has not yet been answered. More broadly, the distinction between “international” terrorism and “domestic” extremism may have made sense when Islamist plotters were assumed to sit in Afghan caves, but is hard to sustain in the age of the home-grown Islamist terrorist and the emergence of a digital world without borders. 34 (f) a refreshed approach to research and innovation, including work with academia and with private sector and international partners to address future challenges. The OIR also envisages that standing arrangements should be instituted to build external expertise and challenge into the future maintenance and development of capabilities, in order to maximise the chance of the CT machine anticipating future developments and reacting to them appropriately. 3.48 The OIR extends to 165 pages, not including Annexes. Some changes are already under way, and others will be initiated soon. In other respects still, the short timescale of the OIR has meant that recommendations lack specificity and will be subject to refinement in the future. Capping document 3.49 The capping document presents the threat picture from both Islamist and XRW terrorism, brings together the various strands of work initiated in the wake of the attacks, with particular emphasis on the strategic themes derived from the postattack reviews, collates the recommendations of the various reviews, and sets out the steps that are envisaged to implement them. 35 4. INDEPENDENT ASSESSMENT Terms of reference 4.1 I was tasked by a letter from the Home Secretary of 28 June 2017 (Annex 4) to produce a restricted-circulation, classified report containing my assessment of the CT Policing and MI5 reviews summarised in the last chapter. 4.2 The letter read, in relevant part: “Thank you for agreeing to provide independent assurance of these internal reviews, to give me and the rest of the National Security Council confidence that all the relevant questions in relation to what was known and done before the attacks have been addressed, and the appropriate conclusions drawn.45 You will be given access to the police and MI5 reviews and any further relevant information you wish to see. You will be able to discuss the material with the agencies involved and ask for follow up information. You should complete your assessment of the police and MI5 reviews by the end of October and provide a summary of your conclusions to me, copied to the Prime Minister and the Intelligence and Security Committee of Parliament.” In accordance with that instruction, I have produced a short classified assessment and conveyed it, together with this open report, to the Prime Minister, the Home Secretary and the ISC. 4.3 The letter continued: “You may also provide a version of your conclusions for public use, which will be published subject to discussions with the relevant information owners and consent of the respective coroners conducting the inquests. This is to ensure that nothing in the public facing documents could prejudice national security or the inquests.” 4.4 45 I informed the Home Secretary that I proposed to exercise my discretion to produce a version of my conclusions for public use. Indeed my aim has to been to put as much of my assessment as possible in the public domain. This open report was submitted by me in final form on 2 November 2017, having been cleared for publication following discussions with MI5 and CT Policing. The purpose of those discussions was to ensure that the contents of this report are factually accurate and that they do not inadvertently endanger national security or risk prejudicing legal proceedings. To ensure that nothing in the report risks The National Security Council (NSC) is a Cabinet Committee, chaired by the Prime Minister and tasked with overseeing issues related to national security, intelligence coordination and defence strategy. 36 prejudicing coronial proceedings, the report has also been passed to the Chief Coroner, HHJ Mark Lucraft Q.C., for his review.46 General approach 4.5 On a restrictive view of my terms of reference, I might perhaps have limited my “assessment” of the police and MI5 reviews to an appraisal of their final reports, together with any supporting documentation that I asked to see. 4.6 The Home Secretary however asked not just for a verdict on others’ work, but for an “assurance” from me that all relevant questions had been asked, and the appropriate conclusions drawn. That exercise, as it seemed to me, required me to immerse myself in the various reviews from an early stage. 4.7 Evaluating others’ work, as indicated in the Foreword to this report, has been a quite different process from conducting my own external reviews. I remained an independent outsider: but my terms of reference gave me a shared interest with MI5 and the police in ensuring the quality of their output. Accordingly I did not wait to judge the finished product but chose rather to involve myself critically and constructively at all stages of the review process, including attendance at key meetings and commenting on successive drafts. This seemed the best way to increase my understanding of the issues, to assess the quality of the underlying work and to maximise my influence on the end product. Assistance 46 4.8 It was clear to me from the start that I lacked existing knowledge of a number of the areas on which the reviews were likely to touch. As Independent Reviewer of Terrorism Legislation from 2011 to 2017 I acquired a reasonable understanding of CT Policing; but save in certain areas, mostly connected with investigatory powers, I began with little specialised insight into the organisation, systems or processes of MI5. Other elements of the task confronting me – for example, the inspection of underlying documents in order to verify that they were fairly represented in the post-attack reports – were more obviously within my expertise but had the potential to be time-consuming. 4.9 Help was accordingly made available, at my request and for short but valuable periods, by three individuals whose security clearance enabled them to read highly-classified material (a basic prerequisite) and whose knowledge and skills helped to supplement my own: It has not been decided in all cases who will conduct each of the inquests as coroner, and in the circumstances the most appropriate person to review the report generally is the Chief Coroner. 37 (a) Graham Webber, Interim Chief Executive of IPCO. IPCO supports the Investigatory Powers Commissioner, a role created under the Investigatory Powers Act 2016 to replace the three former judicial oversight commissioners, including the Intelligence Services Commissioner and the Interception of Communications Commissioner, both of whom had oversight of MI5. My discussions with him over a period of time were helpful in understanding some of the dynamics involved, and in deciding where challenges were needed or should be pressed home. (b) An IPCO inspector with security experience, who shared with me the burden of reviewing a hard copy of each of the primary documents referred to in the MI5 post-attack reviews, in order to assist me in the task of interpreting them and verifying that their contents were fairly represented in the narrative.47 (c) A former Senior Executive of the Counterterrorism Division of the Australian Security Intelligence Organisation (ASIO), who flew from Australia to spend three days attending meetings, reading draft reports and advising me on MI5’s processes and their proposals for change. The resultant insights, coming from the perspective of a friendly intelligence agency which has been confronted with many of the same issues as MI5, were valuable to me. 47 4.10 During the course of my work on these reviews, I also had the opportunity to speak to more than one non-executive director of MI5. These non-executive directors sit on the Management Board, which is chaired by the Director-General, and bring to the table considerable business experience in the UK and elsewhere, including in sectors which are of particular relevance to MI5. These conversations gave me a useful perspective on the corporate culture of MI5. 4.11 Beyond that I chose to work on my own, declining an offer of administrative assistance from the Home Office and arranging meetings through my clerk in Chambers. We also had access to MI5’s computer system for the purpose of examining relevant documents. In the time available it was not possible to perform a similar exercise on the police narratives, which were finalised at a late stage of the process and drawn from a very wide range of sources which were not made available to me in an easily accessible form. I have however examined a number of documents that appear to be of particular importance, and satisfied myself that they are fairly represented in the police narratives. Generally, I observed both the police and the MI5 review teams to be fair-minded and keen to reflect my comments in their work. 38 Nature of my work Working methods 4.12 I was offered and accepted a security pass that allowed me to roam freely within MI5, and the use of a room there where I was able securely to read documents, speak to personnel and write this report. I was also a frequent visitor to New Scotland Yard, where similar facilities were made available to me as required. 4.13 To get a measure of the task ahead, I read public and/or classified versions of past reports on the handling of intelligence prior to a number of other incidents, including the killing of Lee Rigby,48 the Boston Marathon bombing in the United States49 and the Lindt Café siege in Australia.50 4.14 After a joint introduction by MI5 and CT Policing to the intelligence picture on each of the attackers, I had a number of specialised briefings on subjects ranging from the threat (both Islamist and XRW), information, data and compliance, the review of counter-terrorism powers, agent operations, the Northern Ireland experience, the regional counter-terrorism network and MI5’s Behavioural Science Unit. 4.15 At various stages I had the opportunity to speak to the senior management of MI5 (including the Director General, Andrew Parker) and of the police (including the Metropolitan Police Commissioner, Cressida Dick QPM, the Assistant Commissioner and National Lead for CT Policing Mark Rowley QPM and his deputy, Senior National Coordinator, DAC Neil Basu QPM). 4.16 48 49 50 I also spoke to large numbers of individuals with knowledge of the structures and processes described in the various reports, including from the North-West Counter Terrorism Unit which was largely responsible for investigations with relevance to the Manchester attack. I spoke to personnel at all levels of seniority, and read a number of internal MI5 documents, including records of review team interactions with operational decision-makers. But mindful of my limited function, I did not interview or take statements from those who were directly involved in taking decisions on the specific investigations under review. ISC, Report on the intelligence relating to the murder of Fusilier Lee Rigby, HC 795, November 2014. Unclassified summary of information handling and sharing prior to the April 15, 2013 Boston Marathon Bombings, prepared by the Inspectors General of the Intelligence Community, Central Intelligence Agency, Department of Justice, Department of Homeland Security, April 2014. One such report is Martin Place Siege: Joint Commonwealth – New South Wales review, January 2015. Lacking sufficient command of the German language, I was unable to read in addition what appears from press reports to have been a critical assessment – published as my work was drawing to a close – on the actions of the Berlin authorities prior to the Christmas Market vehicle attack of December 2016. 39 Post-attack reviews 4.17 Between July and October I read and commented on several drafts of each of the three MI5 and the four police post-attack reviews. I fed back comments, ranging from the fundamental to the minor. In conjunction with an official from IPCO, I verified that the MI5 reports properly conveyed the significance of the underlying documents, and suggested corrections on the rare occasions when they did not. 4.18 Of particular importance was the opportunity to attend the thorough “pause point” meetings hosted at MI5 and referred to at 3.10 above. At these meetings I observed investigators, agent handlers, desk officers, analysts and police, coordinated by review team leaders, discussing in a frank and often impressively self-critical manner the decisions that were taken at each stage, the “pause points” that could have changed the course of each investigation, and the learning points to be drawn from the sequence of events. Operational improvement review 4.19 On the OIR, equivalent insight was provided by regular meetings of the OIRSB, the high-level group referred to at 3.33 above. The OIRSB was tasked with scoping out the OIR, and as part of that process conducted “deep dives” into a number of the specific issues identified in the OIR’s terms of reference, such as data, closed SOIs and resourcing. Discussions at OIRSB were vigorous and at times close-fought. They gave me an insight, unusual for any outsider, into the heart of debates within the UK intelligence community over counter-terrorism and security. Given the constraints of the short period available for discussion, these meetings struck me as a robust and effective mechanism for formulating and testing the recommendations of the OIR. 4.20 I reviewed and commented on two drafts of the OIR prior to being handed the final version on 17 October. Capping document 4.21 The capping document was the last of the nine reports to be produced, in what appeared to be an intensive process of consultation between MI5 and CT Policing. I did not see it in draft, and was presented with a final copy on 19 October. Extent of my influence 4.22 I formed a positive impression of the integrity of the review teams both at MI5 and CT Policing, and found most of the work with which I was presented, even at an early stage of the process, to be of a good standard. But given the request for assurance in my letter of instruction, it was necessary to test the product as rigorously as I could, and where possible to suggest improvements. 40 4.23 Accordingly, on what must have been (in total) many hundreds of occasions I made specific comments on drafts, asked for proof of assertions, requested documents and briefings, identified issues to be confronted, asked for more thorough accounts, suggested the restructuring of reports, challenged assertions that errors were inconsequential, advised that sensitive material was relevant, discouraged complacency and generally sought to promote the value of selfcriticism. On a limited number of issues I also made the case, sometimes forcefully, for the consideration of specific operational improvements or for furtherreaching recommendations than previous drafts had been prepared to contemplate. 4.24 Some of my suggestions or comments precipitated vigorous discussions, some were more appealing to MI5 than to the police or vice versa, and one or two proved controversial. But all were received with courtesy, many were taken up with enthusiasm, and every one was given effect wholly or in substantial part. 4.25 It would be wrong to overclaim my own influence, given my limited practical knowledge of the systems and processes whose improvement is being discussed, and the fact that the time that I was able to devote to these reviews amounted to only a tiny proportion of that spent, collectively, by those conducting them. But I hope that in many small matters and in one or two broader respects, I was instrumental in pushing things forward; and that my involvement may help give some objective assurance as to the quality of the underlying work. 41 5. CONCLUSION Significance of the internal reviews 5.1 A sequence of four deadly attacks in three months would have been shocking to anyone who had allowed themselves, on the basis of past experience, to believe that the UK’s defences were secure even against the currently high level of threat. But shocks can also be learning points – and this one has been. 5.2 The UK’s CT effort has been effective over the years, and the strong partnership between police and MI5 is widely envied around the world. But even good relationships need to be worked on, and it can sometimes take a crisis to see what is required. These reviews provided the vehicle for that process. Appropriately, they were focused not on structures – the allocation of responsibilities to agencies and police, or the organisation of the regional CT Network – but on operational issues. 5.3 It has been known for large institutions to react to reviews by going through the motions, or digging defensive redoubts. Even when a window is opened to change, it can close again before long. If such temptations existed on this occasion, they have up to this point been successfully resisted. Intelligence professionals and senior police officers have looked at established systems and processes with a questioning eye. They have pushed each other, and persuaded themselves, to contemplate significant change. 5.4 The lasting impact of the OIR, and the other recommendations for future change, will depend on their effective implementation. In some cases, more work will be needed before the destination becomes fully visible. But of particular importance are the three step changes identified at 3.38-3.46 above. As to these: (a) There is certainly room for improvement in analysing and sharing data, and private sector cooperation is vital. (b) The decision to allow intelligence-derived knowledge to be shared more widely beyond intelligence circles marks a significant cultural change. It will allow local police and other agencies to be alerted to threats of which they are currently unaware, and should increase the options for addressing those threats. (c) With an XRW organisation recently proscribed for the first time, there is a strong case – I would say an unanswerable one – for making available the same CT assessment and investigatory tools across the board. 5.5 A frequent underlying theme of discussions was the fact that MI5 and the police bring different but complementary skills to bear in their CT work. MI5 excels in 42 the analysis of intelligence, and the police in criminal investigations. The combination of skills is seen to advantage in the IHM (see 1.19 above), where different attitudes and experience are brought together in the assessment of leads. The OIR helpfully explores means of bringing that joint expertise into play in a wider range of circumstances, particularly where closed SOIs are concerned. 5.6 The OIR required MI5 and CT Policing to consider any legal or ethical questions associated in particular with the issue of sharing data. In conjunction with MI5 lawyers and officials, I have sought to identify the principal legal and ethical issues that are engaged by the proposed reforms, in this and in other areas of the OIR. These issues lie, in large part, within the areas of responsibility already conferred upon the Investigatory Powers Commissioner by the Investigatory Powers Act 2016. To the extent that they do not, I am assured that they are under careful consideration, with legal advisers where appropriate. Nonetheless, I have brought them to the attention of the Home Secretary and of the ISC in my closed report as an extra assurance that they are kept well in mind as part of the implementation process. I have no doubt that the Investigatory Powers Commissioner (and his office, IPCO) will be kept abreast of that process as required. Quality of the internal reviews 5.7 The quality of the reviews was determined, first of all, by the calibre of the officers and staff who were entrusted to produce them. 5.8 I wrote in my final report as Independent Reviewer of Terrorism Legislation that based on my own observations over six years: “.. the hostile narrative of power-hungry security services, police insensitivity to community concerns and laws constantly being ratcheted up to new levels of oppression is, quite simply, false”.51 The character of an institution can quickly change: continued vigilance is called for on the part of all those exercising oversight; and as little as possible should be taken on trust. But my prolonged exposure to both MI5 and to CT Policing in the context of these reviews confirms me in my comment of last December. Despite their daily exposure to a world of profound unpleasantness, both appear firmly rooted in the more positive values of the community they serve. MI5’s young staff, in particular, are more distant than may be easily imagined from the grey and grizzled stereotype of the internal security apparatus. 5.9 51 I never had cause to doubt the integrity of any of the review teams, or of the senior leadership to whom they reported. On both sides of the fence they provided appropriate briefings, often on their own initiative, and offered me full cooperation, David Anderson, The Terrorism Acts in 2015, December 2016, 11.14. 43 whether by inviting me to sensitive internal meetings, volunteering the presence of difficult issues, or responding thoroughly and frankly to my questions. 5.10 The MI5 review teams organised their daunting tasks with skill, adapting to major new demands (including from me) without a fall-off in quality. The post-attack review narratives, when scrutinised against the underlying documents, stand up as impressively thorough and fair. The frankness and openness to criticism that was displayed by staff in post-attack review meetings was encouraging, even if it did not always translate into the occasionally defensive tone of the comment sections in the reviews. 5.11 Led with energy and conspicuous fair-mindedness by ACC Terri Nicholson, the CT Policing review team was equally frank with me, alerting me to possible errors by the police as well as others. The team cheerfully accepted in turn my criticisms and suggestions for improvement to reports whose evolution was not always easy, as officers faced the challenge of integrating different sources of information and formulating learning points to a short timescale. Any undue defensiveness that I encountered from the police was on isolated points, and not from the review team. While I was not able to verify in every respect the accuracy of the police postattack reports against the underlying sources, I believe that where the fundamentals are concerned, they fairly convey the police role in the relevant investigations. 5.12 I found MI5 to be protective of its staff and (like security and intelligence agencies all over the world) inherently cautious in relation to the release of information. Occasionally this led to differences of opinion with CT Policing, or with me, as regards the content or phrasing of the reports. It spoke well for the depth and quality of the working relationship between MI5 and the police that although they bring different qualities and experiences to the table, their discussions always resulted in what I considered to be workable and sensible solutions. 5.13 The OIR is an impressive piece of prescriptive analysis, well reflecting the varied viewpoints and discussions that accompanied its formulation. Not every recommendation has yet solidified; and on the police side, some useful learning points from the post-attack reviews came through rather late in the day, with the consequence that they have not been fully integrated into the OIR. The capping document has however been used to bring all the recommendations together, though not to a consistent template, and has enabled all points of importance to be made. 44 The assessments I am asked to give Were the right questions asked? 5.14 To quote my letter of appointment (Annex 4), I can confirm that “the relevant questions in relation to what was known and done before the attacks” have been thoroughly addressed in the post-attack reviews. Were the right answers given? 5.15 The letter also asked me, more ambitiously, to confirm that “the appropriate conclusions” (emphasis added) have been drawn. 5.16 The processes followed by the review teams seemed to me appropriate, given the relatively short period in which the work had to be done, and I was impressed by the dedication, honesty and expertise of those conducting the reviews. But to give an unqualified endorsement of each of their conclusions would require a degree of operational expertise to which I cannot lay claim. Two points, in particular, occur in this regard. 5.17 First, in relation to the question of whether the Manchester attack could or should have been stopped (2.37 and 3.15 above), I offer no view on whether the re-opening of an investigation into Salman Abedi in early 2017 would, as MI5 concludes “on the clear balance of professional opinion”, have been unlikely to result in the pre-emption of the gathering plot. While that may be right, I prefer to emphasise my agreement with the other point made in this connection: that “there is a high degree of inherent uncertainty in speculating as to what might or might not have been discovered”. 5.18 Secondly, there are recommendations in the OIR whose detailed evaluation requires knowledge and experience rather different from mine. To take one example, a number of the recommendations (relating for example to use of data, to multi-agency engagement and to research and innovation) are likely to have as their effect an increase in the volume of leads. While this may be desirable in principle, the processing of more leads will logically require the transfer of resources from other activities, perhaps including priority investigations. Whether the quality of the extra leads will be such as to justify removing those resources from other areas of MI5’s work is something of an imponderable, at least for me. While I have no reason to distrust the judgements made by the senior leadership of MI5 and CT Policing on this issue, it is ultimately a question of operational feel on which my independent verdict would have little value. 5.19 I would not wish these qualifications to be taken as criticisms of the internal reviews. None of their findings or recommendations strike me as obviously 45 inappropriate. Indeed taken as a whole, the recommendations of the internal reviews appear well-founded and are likely to contribute positively to the UK’s counter-terrorism effort. Will the internal reviews make us safer? 5.20 The terms of reference of the OIR invite consideration of “the overall level of assurance it is possible to provide in this area”.52 Though not the subject of explicit analysis in the OIR itself, it is a question that throws all the others into relief, including the question of whether these reviews have the capacity to make the public safer. 5.21 I conclude with four reflections on this theme, informed by my study of the attacks under review. 5.22 First, intelligence is always imperfect. Commentary after some of the attacks was to the effect that the police or MI5 had been warned that an attacker was dangerous, and that they must therefore have failed in their duty. Tip-offs, whether from friends, family or members of the public, are indeed a vital element of successful intelligence and policing. But we live in a country in which personal freedom is highly valued; in which the thresholds for authorising covert and disruptive action, for arrest and for criminal charge are properly high; and in which people cannot be convicted of serious crime without a randomly-selected jury being sure of their guilt. One piece of intelligence may be helpful in reaching those thresholds but will often be of limited use, particularly when the source is of unknown reliability. 5.23 The Director General of MI5 recently described the work of his staff in the following terms: “They are constantly making tough professional judgments based on fragments of intelligence: pin pricks of light against a dark and shifting canvas.”53 After immersing myself in the minutiae of these investigations, that strikes me as an accurate description of MI5’s counter-terrorism work. The reason why the judgements can be “tough” is that they are made against a background of imperfect information, and yet frequently require staff to choose which of a number of current and potentially deadly threats is most deserving of scarce investigative resource. 52 53 Operational Improvement Review, Terms of Reference (Annex 3). Speech of 17 October 2017 in central London (available on MI5 website). 46 5.24 Secondly, not everything can be stopped. Since we do not live in a surveillance state, or anything approaching it, determined attackers will always have a chance of getting through. The attacks under review illustrate this in a number of ways: (a) Persons heedless of their own lives who would have been far from the top of any rationally-compiled threat grid are always liable to emerge and strike opportunistically with readily-available weapons, as at Westminster and Finsbury Park. (b) A proportion of the tough judgments described by the Director General, understandable as they may have been at the time, will prove in retrospect to have been wrong ones: Manchester provides examples of this. (c) The unpalatable lesson of London Bridge is that even priority subjects of interest in respect of whom sound decisions are being made by MI5 and police may retain the ability to conceal their attack-planning from the authorities. 5.25 It follows that given the current threat, current CT resourcing of around £3 billion per year,54 and the inherent limitations of intelligence work in a free society, no responsible person could offer a copper-bottomed assurance that terrorists will always be stopped. The same is of course true in relation to other types of crime. 54 55 5.26 But thirdly, there is no cause for despair. Most attacks continue to be successfully disrupted, often after intensive and painstaking work, with successful prosecutions and long prison sentences a regular occurrence. The 20 recently thwarted attacks on which I have been briefed, six of which have already ended with sentences of life imprisonment and with many more trials in the pipeline, are evidence of that.55 5.27 Nor could it be said of the attacks under review, save in the case of Finsbury Park, that MI5 and the police were entirely blindsided. Khalid Masood (Westminster) and Salman Abedi (Manchester) had both been subjects of interest, and Khuram Butt (London Bridge) remained under active investigation. Substantial and appropriate coverage was in place around key individuals, and mechanisms designed to assess risk were working as intended. MI5 and CT Policing got a great deal right: At the autumn Spending Review of November 2015, it was announced that cross-government spending on counter-terrorism would be set at £15.1 billion for the period 2016-21. This was an increase of 30% over the figure previously announced and includes substantial new investment in increasing the counter-terrorism capability of the Security and Intelligence Agencies, including MI5. CT Policing received an uplift of 10% over the same period (excluding one-off costs for armed policing and CBRN equipment), but the indicative profile of their grant allocation over the next three years sees a reduction of 7.2% in their budgets. The £15.1 billion figure does not include the full cost of prosecutions or of capabilities such as border security, which serve a range of functions beyond counter-terrorism. See 1.11 above and Annex 2 below. 47 particularly in the case of Manchester, they could have succeeded had the cards fallen differently. 5.28 Fourthly and finally, even marginal improvements are capable of paying dividends. Implementation of the various recommendations cannot guarantee that future plots similar to those under review will be thwarted. But in an increasingly high-volume business, where success and failure depend on tiny margins, there will almost certainly be future cases in which these recommendations will tip the balance in favour of the security forces. MI5 and the police may not stop every attack, in other words, but will be strengthened in their existing ability to stop most of them. 5.29 To conclude, I welcome these careful and trustworthy reviews, and endorse so far as I am qualified to do so their conclusions and their recommendations. 48 ANNEXES ANNEX 1: LIST OF ACRONYMS ACPO Association of Chief Police Officers (now replaced) ALM Al-Muhajiroun (a proscribed organisation) ASIO Australian Security Intelligence Organisation CPS Crown Prosecution Service CT Counter-Terrorism CTP-NOC Counter-Terrorism Policing National Operations Centre CTU Counter-Terrorism Unit CTIU Counter-Terrorism and Intelligence Unit DWP Department for Work and Pensions EEA European Economic Area GCHQ Government Communications Headquarters: UK digital intelligence agency IPCO Investigatory Powers Commissioner’s Office IS, ISIL, ISIS Daesh, the so-called Islamic State in Iraq and Syria JIC Joint Intelligence Committee JTAC Joint Terrorism Analysis Centre MI5 Security Service: UK domestic intelligence agency MI6 Secret Intelligence Service or SIS: UK overseas intelligence agency NCTPHQ National Counter-Terrorism Policing Headquarters NDEU National Domestic Extremism Unit (now replaced) NSC National Security Council OIR Operational Improvement Review OIRSB Operational Improvement Review Steering Board SDSR Strategic Defence and Security Review SIS II Schengen Information System (EU: second generation) SOI Subject of Interest UKBA United Kingdom Borders Agency (now replaced) XRW Extreme Right Wing 50 ANNEX 2:  RECENT CONVICTIONS FOR ISLAMIST ATTACK PLOTS In 2014, Brusthom Ziamani was found in possession of a knife and a hammer. He was arrested, charged under section 5 of the Terrorism Act 2006 (preparing an act of terrorism), convicted in 2015 and sentenced to life with a minimum of 22 years, reduced to 19 years in 2016.  Tarik Hassane and his associate Suheib Majeed plotted to conduct at least one attack using a silenced firearm in 2014. Hassane pleaded guilty to conspiracy to murder and preparing an act of terrorism, and was sentenced in 2016 to life imprisonment with a minimum term of 21 years. Majeed was convicted of the same offences and sentenced to life with a minimum term of 20 years.  Nadir Ali Syed was arrested in November 2014 for his suspected involvement in a plot to behead a member or members of the public. Syed and a number of his associates were charged with preparing an act of terrorism. Syed was sentenced to life in 2016 with a minimum term of 15 years.  Junead Khan devised a plot to attack US military personnel at a UK RAF base. Khan was arrested in July 2015, convicted of preparing an act of terrorism and sentenced in 2016 to life imprisonment with a 12 year minimum sentence.  Haroon Ali Syed attempted to procure an explosive device and/or firearms for an attack in the UK after being inspired by Daesh. Haroon was arrested in September 2016 and after a guilty plea was sentenced in July 2017 to life imprisonment for preparing an act of terrorism with a minimum sentence of 15 years.  Five Birmingham-based individuals were arrested following the recovery of assorted weapons from a vehicle attributed to the group. Three former TACT prisoners Khobaib Hussain, Naweed Mahmood Ali and Mohibur Rahman, together with Tahir Aziz, were subsequently charged with preparing acts of terrorism and were sentenced in August 2017 to life imprisonment with minimum terms of between 15 and 20 years. Further detail on these and other terrorism convictions may be found on the website of the CounterTerrorism Division of the CPS. 51 ANNEX 3: OIR TERMS OF REFERENCE MI5/Police Operational Improvement Review: Managing the Lifecycle of Counter Terrorism Investigations DG MI5 and the Commissioner of the Metropolitan Police have commissioned a short review to consider how to optimise processes for managing the lifecycle of counter terrorism investigations in the context of the changing terrorist threat. The Review will focus on identifying any further improvements that can be made to the processes for triaging incoming intelligence (‘leads’), prioritising investigations, and managing the residual risk posed by individuals after investigations are closed. The aim of this Review will be to: a) Further improve operational effectiveness, building on the strong counter terrorism machinery that has been built in the UK over more than a decade. Given the short timescales, this is likely to be through recommending areas for further work to be taken forward following completion of the Review. b) Provide MI5 and the Metropolitan Police’s input on this issue into any wider reviews of the UK’s counter terrorist and counter extremism systems. Scope The Review will consider:  Whether there are any further improvements that can be made in how leads, prioritisation and triage processes operate;  The process by which individuals are categorised as a closed subject of interest, and how cases are then reviewed and escalated where indicators of potential re engagement in terrorist related activity are identified;  What data sources, tools and approaches, both tactical and strategic can best support this work;  Which partners can contribute, and how we can further strengthen joint working in managing this risk;  How data is shared and links are made with other organisations / interventions (such as PREVENT or community policing), where appropriate;  Any policy, legal or ethical questions associated with this issue that require consideration;  The overall level of assurance it is possible to provide in this area, and the resource and prioritisation challenges associated with managing this risk. The Review will liaise with the Legacy reviews on the Westminster, Manchester, London Bridge and Finsbury Park (Police only) attacks to ensure that their learning is reflected where it touches on the issues above. 52 The Review is not intended to examine the entire investigative process or to replicate broader issues being considered in reviews of CONTEST or CT legislation:  The OIR will examine the definition of Priority 1–4 investigations only in so far as it relates to leads, triage and the opening and closure of SOI records.  The OIR will not: o make recommendations on thresholds for and allocation of intrusive resources, or the development of capability and processes in Priority investigations; o examine the suspension of Priority investigations where resources (including investigators) are required for more urgent investigations; o examine the balance of P1-4 investigative casework across Thames House and the Regions. o Review the processes for managing P1-4 investigative casework (eg ELGs and JOTS) - Report The Review will seek to identify and begin to implement changes as areas for further improvement are identified. A final report will be produced within approximately three months in CLOSED, with an OPEN summary to be published and/or provided to Reviews on wider CT processes, as appropriate. Any outcomes produced will of course be disclosable in inquests and legal proceedings. Resources The Review will be led by [MI5’s Head of CT Policy and Strategy],56 and ACC Terri Nicholson (Deputy Senior National Coordinator), Met Police and supported by a Review team comprised of: - 2 x staff from MI5 central policy team, with investigative background, and ensuring join up with wider MI5 work on addressing future challenges and threats. Policing representative HO representatives GCHQ representative The work will be overseen by a Director level Steering Group, which will meet every fortnight and will include: - MI5 and Police co-chairs - Director Analysis, MI5 - Director Policy, MI5 - Behavioural Science Unit, MI5 - Director Intelligence, NCA - Deputy Director Counter Terrorism, SIS - Director National Security, Home Office - Director Prevent, Home Office - Director Capability, GCHQ 56 Author’s note: square brackets here and below indicate the redaction of names or other identifiers and, in this case, the insertion of a job title instead. 53 - [ ] MI5 Non-Executive Director [ ] Chief Scientific Advisor, MI5 & SIS The Review will also consult with a number of independent individuals with expertise in data science and risk, as well as with industry partners, OGDs with big data expertise, think tanks and international partners. 54 ANNEX 4: MY LETTER OF APPOINTMENT 1" p] Heme Secretary 2 Marsham Street Landcn SW1P 4UP vmyige c- Wednesday ES June 2017' Dear David, INDEPENDENT ASSURANCE OF THE MID Ml?- REVIEWS THE LONDON MID MANCHESTER TERRORIST ATTACKS Fcrur tenprist attacks have eccurred in the UH: between March and June 201 T, in Westminster. Manchester, Lendpn Bridge and Finsbury Parlt- As is usual practice after any machr incident, MIE and the Hatipnal Caunter Terrpiism Felice Headquarters have already launched internal reviews that will. ameng ptherthings: I censider what was abeut the attackers and any alleged cc:- canspiraters plicir te the attacks taking place: I review assessments and decisians made pnertp the attacks taking place, in relatian tp that intelligence; and I identityr whetherthe pmcesses and systems that supperted these decisicins {including fer exam ple the management at ctesed subjects at interest ?feniier subjects [if interest in tenprism investigatidnsj are apprcipriate tci the evetving threat, and new they can be impreved- Thanlt yp-u ipr agreeing tp previde independent assurance at these internal reviews. te give me and the rest at the Natipnal Security Cduncil cantidence that all the relevant questions in relatien te what was Itan and dcne befere the attacks have been addressed. and the apprepliate drawn- ?r'eu will be given access tn the pclice and MIE reviews and any further relevant inticin'natipn ypu wish tp see. ?fciu will be able ta discuss the matelial with the agencies and astt fer fallew up inferrnatien- ?r'eu sheuld cemplete ypur assessment at the pplice and MIE reviews by the end at Dctcib-er and previde a summary cii ydur canclusidns ta me, cdpied tn the Pn me Minister and the Intelligence and Security Sammittee ei Partiarnent. ?r'eu may alsp previde a versian ei yaur canclusipns fer public use, which will be published subject tci discussiens with the relevant infennatipn dimers and cansent di 55 the respective cerenere cenducti ng the inqueete- This is eneure it'iet in the public tecing document ceulci prejudice netienel security er the inqueste- Thenlt yeu for agreeing te undertake this iguOLM Rt Hun Amber Rudd MP 56 ANNEX 5: MI5 INVESTIGATIVE PROCESSES This Annex replicates the information on MI5 investigative processes included as an annex to the ISC’s public report on the intelligence relating to the murder of Fusilier Lee Rigby. It has been subject to minor updates to reflect changes to these processes since the ISC reported in November 2014. The following provides a brief overview of MI5’s investigative approach, drawing particular attention to the points at which there is challenge (both internally and from Whitehall) and where there is flexibility to respond to developments. This document is not an exhaustive explanation of our prioritisation processes. We would like to provide you with more detail on this during briefing sessions. Internal Processes The rise in the Islamist extremist threat necessitated a response akin to an industrialisation of MI5’s approach to investigation. Since 2011, we have in place a formal triage process for incoming threat intelligence, a prioritisation system which is revisited regularly for adjustments according to the waxing and waning of risk and a higher level review process to set strategic priorities. • On receipt, new intelligence is tested for links to existing investigations and forwarded to the appropriate team where those links exist. Alternatively, where intelligence does not relate to existing investigations it is assessed for the nature and extent of national security risk, credibility, actionability and proportionality and a new lead or investigation is launched if appropriate.  A Lead is the term to describe all intelligence or information that is not linked to an ongoing investigation that, following initial assessment, suggests activities of National Security (NS) interest requiring investigation by MI5 and CT Police.  A trace is a request for a check across MI5 databases to determine whether MI5 holds adverse information or to check that an individual is not known to MI5 already. A liaison trace is a request from a liaison partner, such as a foreign intelligence service, to determine if we hold any information relevant to their investigation(s). • Investigations are given a priority according to the risk they carry. The broad categories are described in the table below. This table does not attempt to explain in detail our prioritisation process but is designed to provide an overview of the way we manage our investigations. There are no stringent rules for what resources should be given to a particular investigation and actions are taken based on whether it is judged necessary and proportionate to do so and on the balance of risk in other investigations. The priority level is regularly tested at senior management level, and priority levels are altered as changes are noted in the activities or aspirations of the individuals or networks we are investigating. 57 Category Definition for Islamist terrorism Investigations Priority 1 – Attack Planning Investigations into individuals or networks where there is credible and actionable intelligence of significant (P1a) or smaller scale (P1b) attack planning. Priority 2 – High (linked to Investigation into individuals or networks where there is attack planning) and medium involvement in activities including: risk (not directly linked to attack  a serious intent to travel overseas to undertake planning) extremist activity. fighting with an extremist group (P2H) Priority 3 – Uncorroborated intelligence relating to any extremist activity. Priority 4 - Risk of reengagement following disruption of involvement in extremist activity.  large scale fundraising (P2H)  significant terrorist training (P2H)  supply of false documents (P2M)  smaller scale fundraising (P2M). Investigations or networks that require further action to determine whether they pose a threat. Investigations into individuals who have previously posed a serious threat to national security but who we judge are not currently involved in such activities but there is a risk of re-engagement. For instance, release of individuals imprisoned for terrorism offences. • Within most investigations we also prioritise the subjects of interest we investigate. This is done through the allocation of ‘Tiers’ to SOIs. The Tier of an SOI within an investigation can change regularly depending on the importance of that individual. Tiers reflect the position/importance of SOIs within the investigation that they are assigned to. This will help investigators manage their targets and support understanding of the investigation. • Tier 1: Main targets of an investigation – targets will likely be involved in all aspects of the activities under investigation. • Tier 2: Key contacts of the main targets – targets will likely be involved in a significant portion of the activities under investigation. • Tier 3: Contact of Tier 1 and Tier 2 targets – targets will likely be involved in only marginal aspects of the activities under investigation. 58 • Sitting within our investigative structure we have a Strategic Intelligence Group, specifically designed to provide assessments which inform resource allocation decisions and challenge the assumptions of investigators. • Every week, the head of investigations reviews intelligence developments in a formal meeting, incorporating updates from those leading individual operations, input from police and SIA colleagues and an analytical feed from JTAC. This process results in the production of the highest risk investigations, the apportionment of resources accordingly and the weekly letter we send to the PUS at the Home Office [NB: this is now sent to the Home Secretary]. • Also every week, the ICT senior management team considers a weekly dashboard of wider resourcing issues, such as the total number of investigations, staffing levels, the processing of leads and any backlog thereof. This dashboard of management information allows for the flexible reinforcement of staff and other resources where the need arises. • The Director General (DG) is briefed on a weekly basis as to the main developments and risks. The Deputy Director General has oversight of the use of intrusive investigation measures before they are sent to the Home Secretary for consideration. • Every quarter, there is a thorough review of all our ICT casework. Some investigations are closed, others are selected for an injection of resource. Out of this quarterly process also comes an internal report on trends in our casework, which informs a more strategic review of our investigative footprint. At this point the ICT business also feeds into the MI5’s Quarterly Performance Report (of which more below). • Every week we brief ACSO on key developments in our investigations. Separately, the Executive Liaison Group (ELG) process exists to allow us to jointly agree with police the management of risk where we identify a risk to the public from our investigations. At working level a police senior investigator is appointed to major MI5 investigations and is an integral part of our management team making decisions on resourcing and priorities. • Similarly, the Director of ICT agrees monthly with his counterparts in SIS and GCHQ any strategic shifts required to improve our collective response to developments in the threat. The head of JTAC also sits on this body. • These SIA CT heads also agree a joint annual business plan for ICT. External Visibility and Scrutiny The processes outlined above have a number of docking points with Ministers and senior officials to facilitate scrutiny and challenge to the emphasis of our CT effort, whilst preserving our operational independence to take case-specific decisions. Not all of the portals below are specific to ICT, but naturally our single largest area of business features prominently. • Principal among these at a strategic level are, of course, the Home Secretary’s Weekly Security Meeting and the National Security Council and its subject specific sub-committees. • Beneath these structures, our priorities are discussed and scrutinised across government via the CONTEST provisions. Key among these elements are the PURSUE Board, the Overseas CONTEST Group and the bi-annual CONTEST Performance Report. • Our quarterly review process forms part of the MI5 Quarterly Performance Report, which we share with Home Office (Office for Security and Counter Terrorism) colleagues. We also write our own quarterly report on trends in our investigations, and share that with colleagues in 59 Whitehall. Additionally, we supply intelligence and statistical data to inform JTAC’s quarterly review. • The DG attends the Weekly Security Meeting at the Home Office, which is chaired by the Home Secretary and discusses the highest priority cases. The DG also has regular bilateral meetings with the National Security Adviser, the PUS at the Home Office, and the Director General of the National Crime Agency, and sees the Prime Minister on an ad hoc basis. • At a more granular level of detail, there is the DG’s weekly CT letter to the Home Secretary highlighting significant developments in our highest priority investigations for the week ahead. • MI5 ICT senior managers also engage in frequent dialogue with counterparts in OSCT on matters relating to warrantry and disruptive measures, such as TPIMs and deprivation of nationality. Leads The management of all new CT Lead intelligence and threat reporting not linked to ongoing CT investigations received by both MI5 and the Police is conducted through the Intelligence Handling Model (IHM). This is a joint initiative between MI5 and the Police and provides a single point of entry for intelligence and ensures new leads benefit, where appropriate, from a co-ordinated MI5, GCHQ, JTAC and CT Police tracing and expertise. This co-ordination takes place by dedicated teams in MI5. The IHM provides a robust framework to ensure that finite covert investigative resources are directed against the most credible new leads – and that leads lacking credibility are resolved in the most appropriate way, without significant covert investigative resource. A Lead is the term to describe all intelligence or information that is not linked to an ongoing investigation that, following initial assessment, suggests activities of National Security (NS) concern. An Investigation is a Lead that has met the threshold for significant covert resource to be deployed. Investigations will make use of the full range of covert investigative actions, as necessary. Risk, Credibility, Actionability and Proportionality (RCAP) are the key principles for the assessment and decision-making in the IHM. The RCAP Framework is used at all stages of assessment commencing with the Single Point of Entry (SPOE) and most importantly, at the initial point of triage and assessment. Each lead passes through the four stages of lead development: Receive, Assess, Develop, Decide. I. Receive. All information and intelligence entering the Security Service or Police CT Network is received via a clearly identified and recognised Single Point Of Entry (SPOE) where its receipt is recorded. Processes are in place to receive and assess intelligence 24hrs a day, 365 days a year. II. Assess. Assessment of intelligence occurs at each stage of the process, beginning with the SPOE. All information and intelligence received is assessed to determine if it fulfils the definition of a lead. This is the ‘initial assessment’ referred to in the definition of a lead. Although tracing against indices will usually be sufficient in order to come to a judgement that information or intelligence meets the definition of a lead, it may be that further preliminary actions are necessary, such as a call back to the provider of the information. 60 Identified leads are assessed using the RCAP Framework. They are allocated a grading according to the nature of the reporting and credibility assessment. III. Develop. Lead development is the process of identifying intelligence gaps and requirements, and the further research and actions necessary, to enable a more informed assessment of the lead. All development activities should be necessary and proportionate to the level of risk to national security. Leads should be developed where possible without the application of significant covert resource (such as surveillance or intercept). Any application of resources must be proportionate to the risk held by the lead and considered within the overall prioritisation framework. MI5 and Police endeavour to agree and deploy resources in accordance with the risk and credibility assessment. Each organisation is accountable for the deployment of its own resources. IV. Decide. Decisions on what action is to be taken on a lead occur at each stage of the lead assessment process, beginning with the SPOE. The grading will be reviewed continually and amended where appropriate to ensure it accurately reflects the risk and credibility assessment. 61