Introduction
The four years following the outbreak of the Russo-Ukrainian War on 24 February 2022 have seen a series of profound changes to Sweden’s defence, foreign, and security policies. On 28 February 2022, a mere four days after the outbreak of the war, the Swedish parliament decided to revise the state budget to allow a donation of helmets, body armour, food, and 5,000 anti-tank missiles to Ukraine worth 413.5 million kronor – a contribution followed by no fewer than 21 further donations to a total value of some 100 billion kronor to March 2026 (Sveriges Riksdag, 2022b).1
More followed. On 7 April 2022 and, notably, on 17 December 2024, when the defence budget for 2025–2030 was decided, the Riksdag approved significant increases to the budgetary allocations to territorial defence, calling time on Sweden’s roughly 25-year-old policy of participating in international armed missions far beyond its borders in places such as Afghanistan, Mali, and the waters off Somalia (Sveriges Riksdag, 2022c, 2024c). On 16 May 2022, Sweden applied for membership of NATO. The application was ratified by the riksdag on 23 March 2023; a year later, on 7 March 2024, Sweden joined the alliance (Sveriges Riksdag, 2023).2
And, finally, with the Riksdag’s decisions of 5 December 2024 and 20 March 2025 to commission Swedish armed forces to serve under Nato command in Latvia and Poland, respectively, Sweden’s policy of either turning a blind eye to its Baltic neighbourhood altogether (the approach taken for the entire Cold War) or offering qualified support, primarily in terms of politics (the approach of the early post-Cold War years), was replaced by a policy of engagement – military deterrence directed at the Russian Federation (Sveriges Riksdag, 2024b, 2025b).
With such radical change in the air, it was only to be expected that calls for a major revision of the institutional structure of Sweden’s intelligence would be made sooner or later.3 On 26 October 2023, the Swedish government designated the former prime minister and foreign minister Carl Bildt Special Investigator, tasking him with conducting a review of “the functions of intelligence in support of Sweden’s foreign, security and defence policies, as well as for the assessment of the threats to the security of Sweden” (Dir. 2023:150).4 The Intelligence Inquiry (Underrättelseutredningen) he conducted presented its report, “En reformerad underrättelseverksamhet” [a reformed intelligence service] (SOU 2025:78), around eighteen months later, on 13 June 2025.5
The government, as stated in the Inquiry’s terms of reference, called for an investigation into aspects pertaining both to “defence intelligence” – a term introduced in 2000 with the Riksdag’s approval of a law on such intelligence (Prop. 1999/2000:25; Sveriges Riksdag, 2000) – and counterintelligence in its various forms (e.g., counterespionage, counterterrorism). Although the Inquiry did indeed address both areas, its most radical proposal concerned defence intelligence: that the collection, processing, and analysis of försvarsunderrättelser, to use the Swedish term, including collection through clandestine methods carried out by the Office of Special Collection (Kontoret för särskild inhämtning, KSI), should be transferred from the Military Intelligence and Security Service (Militära underrättelse- och säkerhetstjänsten, MUST), hitherto operating under the aegis of the Commander-in-Chief of the Swedish Armed Forces (Överbefälhavaren, ÖB), to a new and separate government authority: a civilian intelligence service, referred to by the Inquiry as the Foreign Intelligence Agency (Utrikestjänsten).
The Inquiry also proposed that the proper term for the type of intelligence of concern to the new agency should be “foreign intelligence” rather than “defence intelligence”; that the agency should be a “national resource immediately below the Government”; and that it should be “headed by a national Director of Intelligence”. The new authority should be in operation, the Inquiry further advised, no later than January 2027.
While pointing to a host of different reasons justifying the establishment of the new agency (outlined in further depth below), there is little doubt that perhaps the most prominent of these consisted in the Inquiry’s claim that MUST had failed to provide a correct assessment of Russian motives and intentions with regard to Ukraine. As noted by the Inquiry, MUST’s “assessments of the political motives with the Russian leadership fell short” (SOU 2025:78, p. 24); “its analysis of Russia’s inclinations in terms of risk-taking was plain wrong”; and despite access to “information speaking in different directions”, much-needed revisions were not made (SOU 2025:78, p. 24). In summary, the Inquiry described the shortcomings of MUST in terms of an intelligence failure (SOU 2025:78, p. 24).
As the Inquiry noted:
Intelligence failures are hardly unique. There have been many situations in which intelligence organizations failed to see a threat or a change coming. It is in the nature of intelligence to contain a degree of uncertainty, and there is never any guarantee that every assessment will prove correct. What many of these failures have in common, however, is that attempts have been made, in hindsight, to learn from them so as to reduce the likelihood of their recurrence. There is also a relatively rich body of literature and research on the subject. It is important that we do the same. (SOU 2025:78, p. 23)
This article does not seek to establish whether any reorganization of Sweden’s intelligence is necessary, nor to dispute the Inquiry’s proposal to establish a new foreign intelligence agency. Whether such an agency would improve Sweden’s early warning capacity remains an open question. The focus here, rather, is the Inquiry’s representation of the rationale for the establishment of such an agency, or, to be more precise, a specific but arguably very important element of this rationale: the Inquiry’s take on MUST’s performance in the run-up to the Russo-Ukrainian War. The article critically reviews the Inquiry’s attempt to write the history of what it understands to be an intelligence failure: MUST’s incorrect assessment of the Russian government’s intentions towards Ukraine. The conclusion drawn here is that the Inquiry’s almost exclusive focus on the bureaucratic level leaves several crucial questions about the political level unaddressed.
Background
Before addressing the arguments and findings underlying the Inquiry’s conclusion, it should be noted that neither the initiative to have MUST relocated and transformed into a government authority in its own right, nor the claims on its shortcomings in the run-up to the Russo-Ukrainian War, were new. The question regarding a relocation and transformation of MUST was subject to analysis and reflection with an earlier inquiry into Sweden’s intelligence: the Intelligence Inquiry of 1996 headed by Johan Hirschfeldt, Chief Justice of the Swedish Court of Appeal (referred to as the Hirschfeldt Inquiry below),6 which presented its report in March 1999 (SOU 1999:37, pp. 274–276). In contrast to the Intelligence Inquiry of 2025, Hirschfeldt advised against having MUST relocated from ÖB’s supervision and leadership.
Among the reasons cited were, inter alia, that ÖB, in part, would no longer be fully in control of Sweden’s military preparedness; that if the military intelligence services were to be able to identify and analyse political and economic developments and future threats and to provide intelligence for planning and development, they must have the capacity to carry out strategic assessments requiring the establishment of a parallel structure for the collection and analysis of such intelligence; that the establishment of a “multifaceted, national intelligence organ, charged with handling strategic intelligence from a general security perspective” would impair the independence of a range of other government authorities involved in various aspects of national intelligence; and, lastly, that the “activities [of such an agency] would by necessity … become parallel to the activities that are presently being pursued within the Foreign Ministry” (SOU 1999:37, pp. 276–277).
When it comes to the Inquiry’s representation of MUST’s alleged intelligence failure, it should be noted that on 15 June 2022 – almost three years to the day before the presentation of the Inquiry’s report – Mikael Holmström, security policy reporter with the newspaper Dagens Nyheter (DN), presented the results of extensive journalistic work based on information from sources with, by all accounts, good insight into the intelligence process in the run-up to the war. “As late as the weeks before Russia’s attack,” one of the sources held, “MUST indicated that a major attack was not very likely”. There was also no report “indicating an attack of the magnitude that subsequently occurred.” Another source detailed that “our Swedish intelligence service had insisted for months that there would be invasion” and that “we made a complete misjudgement”. Holmström also noted that on 10 January 2022, at the annual Folk och Försvar conference in Sälen, the head of MUST, Major General Lena Hallin, when presenting the agency’s threat assessment, had emphasized that “a major Russian military attack [admittedly] could not be ruled out”, but that it was “not [considered] likely”. Further, in a subsequent interview with DN, Hallin stated that Russia was “trying as far as possible to avoid military intervention” and that “it is a matter of maintaining control over the escalation, not challenging too much”. Other sources referred to by Holmström had stated that MUST’s reports to the government were “in line with [Hallin’s presentation] at the national conference”: that Russia “would not escalate by launching a major military offensive, since that would be too risky and expensive” (Holmström, 2022).
In the article, Holmström, with reference to anonymous sources within the government, also noted that at a meeting with the Organisation for Security and Co-operation in Europe (OSSE) in Stockholm on 2 December 2021, Foreign Minister Ann Linde, speaking to U.S. Secretary of State Anthony Blinken, was said to have questioned American intelligence about an impending Russian attack on Ukraine, arguing that “according to Swedish intelligence assessments the scenario of an invasion had been ruled out”. The same sources also held that Prime Minister Magdalena Andersson – although no specific date was given here – “had been unhappy with the intelligence provided by the Foreign and Defence Ministries” (Holmström, 2022).
The Intelligence Inquiry’s treatment of these two crucially important contributions to the subject under scrutiny must be considered debatable. While noting, in accordance with Hirschfeldt, that depriving ÖB of the responsibility for Sweden’s military preparedness would be in conflict with some fundamental guiding principles of the Armed Forces (SOU 2025:78, pp. 70–71), the Intelligence Inquiry failed to address any of the other key points raised by Hirschfeldt, most importantly, that the activities of such an organ “would by necessity … become parallel to the activities that are presently being pursued within the Foreign Ministry” (SOU 1999:37, pp. 276–277). Moreover, its claim that the Hirschfeldt Inquiry had concluded “that the time was not ripe” for the establishment of a civilian intelligence service must be considered simply misleading (SOU 2025:78, p. 172). There exists no such formulation or line of reasoning in the Hirschfeldt Inquiry report; its conclusion that the establishment of such an organ should be considered inadvisable was, as outlined above, founded on the very specific institutional causes it references.
Second, the Inquiry passed over Holmström’s article entirely. His work is not referred to anywhere in the report. This failure to address it, in a section on earlier investigations, for example, also suggests that the Inquiry did not reflect on the possibility that Holmström’s account of MUST’s alleged misconduct might, over the years, have become “an established truth” among some key officials within the intelligence organs and the Government Offices, serving to define the boundaries of what it was appropriate to say when they were interviewed by the Inquiry.
Besides pointing to MUST’s failure, the Inquiry referred also to a host of additional reasons motivating the establishment of the new intelligence agency, including, among others, a “wider” and “sharper” set of threats against Sweden (SOU 2025:78, pp. 18, 91–96); a blurring of the traditional boundary between foreign intelligence and the domestic security services (SOU 2025:78, p. 18); new opportunities for cooperation with the intelligence agencies of other states arising from Sweden’s membership of the European Union and NATO (SOU 2025:78, pp. 107–118); and “altogether new opportunities for collecting more or less open information” through means such as “commercial actors with capabilities that often match those of the most elaborate government resources (SOU 2025:78, pp. 18–19, 96–104). The Inquiry also pointed to aspects of a more operational nature, including “the increasingly important role played by the Prime Minister with regard to Sweden’s overall security policy”, which has entailed “increased demands on the Prime Minister’s Office (Statsrådsberedningen) to coordinate Sweden’s activities in both bilateral and multilateral institutions for cooperation, primarily the EU and … NATO” (SOU 2025:78, pp. 126–127). The Government’s need for intelligence, the Inquiry asserted, has become more acute as a result (SOU 2025:78, p. 128).
The Inquiry’s reference to MUST’s alleged intelligence failure differs, however, from these reasons in several important respects. First, the very act of pointing to this particular episode required the Inquiry to write an intelligence history of the months preceding the war’s outbreak. Second, this history took the form of a post-mortem analysis of a type often seen in the wake of intelligence failures, and while staying clear of all details that might help us understand the causes of this particular failure, the Inquiry nonetheless drew a host of conclusions as if significant time and analytical efforts had been devoted to having it investigated. As noted by the Inquiry:
It was considered justified, as part of this investigation, to study the period preceding the full-scale Russian aggression against Ukraine. …. We have therefore gone through written reports from the intelligence agencies to the Government and the Government Offices covering an extended period of time. We have also held discussions with relevant parts of the intelligence community and decision-makers in the Government Offices. (SOU 2025:78, pp. 22–23)
However, given the many hours of work that seem to have gone into reading reports and interviewing key staff in the intelligence organs and the Government Offices, and the importance ascribed to the resulting findings in justifying the need for a new intelligence agency, it is surprising that this work was addressed in little more than four or five pages of the Inquiry’s report. As a result, it is remarkably hard to reconstruct how the Inquiry reached its conclusions. Indeed, it seems fair to ask in what sense we can speak of an investigation at all when, by all accounts, the analytical work underlying its conclusions appears never to have been committed to any document.
Here we may point to earlier public inquiries where some aspect of the conduct of Sweden’s military intelligence has been subject to analysis. Examples include the Sandler and Tammelin Inquiries whose respective reports addressed the role of the Sweden’s military intelligence with regard to the stream of Baltic refugees to Sweden in 1943–1945 (the Sandler Inquiry’s fourth report), and into alleged misconduct with the so-called C-byrån of the Defence Staff’s Intelligence Department (the Tammelin Inquiry).7 While classified for several decades, these two Inquiry reports still allowed at least for later generations to consider their lines of reasoning and conclusions
The Intelligence Inquiry’s Representation of MUST’s Alleged Failure
Unlike Holmström, whose account addressed, if briefly, and if setting out to illustrate purportedly harmful effects on diplomacy and decision-making arising from MUST’s purportedly faulty analysis – the Intelligence Inquiry almost exclusively stuck to questions of bureaucracy (or, more precisely, to MUST itself), in detailing both the nature of the alleged intelligence failure and its causes. Three sets of dysfunctions – in analysis, organization and communication – were singled out for discussion.
The key features of the first have already been referred to above – MUST’s assessments of Russian motives were impaired by a number of different weaknesses, most notably by their failure to account for the political motives of the Russian leadership (SOU 2025:78, pp. 23–24, 86, 157). With regard to MUST’s organization, the Inquiry concluded that “MUST in practice fulfils two different tasks – a civilian intelligence task with the government as its main client and a military intelligence and security service task with [the Commander-in-Chief] as its main client”; that overall, this “dual command … poses several challenges for governance and management”; that “in light of the increasingly divergent intelligence needs of the government and the Commander-in Chief, it has proven challenging for MUST to satisfy both”; and that “the new intelligence needs, both military and broader civilian, imply a strong pressure on MUST to change its mission” (SOU 2025:78, p. 155).
The Inquiry further noted that “a high level of security protection has led to a certain degree of compartmentalization within and between the various parts of [MUST]”, and that this has resulted in “a certain degree of inertia due to large distances organizationally, physically and in working methods” (SOU 2025:78, pp. 156–157). In addition, the Inquiry laid down that “information obtained through special methods is valued more highly than aggregated analysis and open information”, which is partly “based on a tendency to base analysis mainly on unique information” (SOU 2025:78, p. 157). This, according to the report, has “limited MUST’s ability to meet the needs of its clients, and the intelligence system as a whole therefore finds it difficult to extract maximum synergies from the resources invested” (SOU 2025:78, p. 157).
Finally, with regard to the alleged shortcomings in MUST’s external communications, the Inquiry concluded that “repeated evaluations of MUST’s intelligence production have identified certain challenges, particularly with regard to recipients in the Government Offices”:
MUST presentations and written reports are seen as cumbersome and unclear in terms of both language and assessments. The uncertainties that always exist in intelligence assessments are not communicated clearly. There are advantages to the long-term integration of military and broader security policy processing and analysis within MUST. However, there is a perception that such a balanced picture is not being presented among the reports reaching the Government and the Riksdag. (SOU 2025:78, p. 157)
While fairly detailed when it came to identifying and describing these dysfunctions, the Inquiry made no attempt of any kind to further analyse the many questions that might be formulated in relation to them. For example: if there truly existed information “pointing in different directions”, as the Inquiry claimed, why did MUST fail to re-evaluate its analysis so it might seek to address a truly comprehensive set of possible scenarios? And in what ways, more exactly, did the alleged organizational shortcomings – the dual command, the compartmentalization, and the tendency to attach greater importance to intelligence collected using special methods – contribute to MUST’s failure to correctly assess Russia’s intentions with regard to Ukraine?
One short remark in the Inquiry’s report provides a clue into what the Inquiry may have found, or at least believed itself to have found, when analysing MUST. When pointing to the benefits of “the new agency”, the Inquiry held that “it should help to reduce the risk of fundamental miscalculations”, and that “the run-up to the large-scale [Russian] invasion in 2022 points to the risks of having too monolithic and insular systems” (SOU 2025:78, p. 27). Remarks along the same line of reasoning were seen when the Inquiry’s report was presented on 13 June 2025. The head of the Inquiry, Carl Bildt, here noted here that
We have conducted a thorough review of the events leading up to the full-scale Russian invasion of Ukraine. It did not turn out quite right, to put it mildly, and we have spent a considerable amount of time on the whole process. How it came about and what lessons can be learned from this still leads to some friction (Government Offices of Sweden, 2025b, 11:14).
Bildt further noted that having both a military intelligence service and a civilian intelligence service brought a positive degree of pluralism to the system:
Perhaps in the run-up to Ukraine, it became clear that it was too much of just one opinion and there was no discussion or evaluation of the various possibilities. We do have others, we have Försvarets forskningsinstitut [the Swedish Defence Research Institute] and others who are involved in the whole thing, but it became a little too monotonous in the system, and because it became too monotonous, there was not sufficient sensitivity when evaluating information other than what we previously thought we had. And then you get a little more, you get a system with a little more intellectual vigour and excitement, and I think we need that in a much more complicated world if we look a few years ahead (Government Offices of Sweden, 2025b, 19:09–19:50).
The third dysfunction, the shortcomings in MUST’s external communications, can be said to concern less the author and sender of the intelligence assessment and more the recipient: the Government and the Government Offices – an observation obliging us to consider the political level in the Intelligence Inquiry’s analysis.
Although, as mentioned, the Inquiry dealt almost exclusively with bureaucratic matters, at least two findings could be said to belong at the political level. The first is that “responsibility for a balanced intelligence assessment in the Swedish system lies with MUST” (SOU 2025:78, p. 23); the second is that “MUST’s ongoing assessments were of great importance to how the government assessed developments” (SOU 2025:78, p. 86). While the Inquiry did not address the question of when and by whom this responsibility was given to MUST, it seems possible, even probable, that such a division of responsibility within the Swedish foreign intelligence service may have taken place in accordance with political directives.8 It is worthwhile noting, also, that the Inquiry did not address the question of why MUST’s “flawed” ongoing assessments were allowed to exert such a significant influence on the government’s view of the risk of a Russian attack, nor where any political responsibility for this might lie.
In short, the Inquiry’s overall picture of the purported intelligence failure thus appears, if only implicitly, to be drawn as a causal chain: MUST, failing analytically, organisationally, and communicatively to live up to the expectations of the special role it had been assigned, namely responsibility for the overall intelligence assessment, led the government astray. This special role of intelligence coordinator was emphasized not only to illustrate the flow of intelligence from the relevant agencies to the government, but also as an important, even decisive, explanation of why the government failed to see what was coming.
Analysis: Three Objections to the Intelligence Inquiry’s Methods
In this section, the Inquiry’s analysis will be addressed from the perspectives of three fields of inquiry concerning intelligence: first, from the perspective of foreign intelligence; second, from that of intelligence failure; and, third, from the politicization of intelligence.
Foreign Intelligence
Regarding the first of these perspectives, it should be noted, to begin with, that the responsibility for a state’s overall assessment of threats and risks in international politics necessarily lies at the political level – with the Prime Minister and with the heads of the ministries most directly concerned, primarily the Ministry of Defence, the Ministry of Justice and the Ministry for Foreign Affairs. It should also be noted here that this is, as mentioned above, a point the Inquiry itself made when itemizing a new foreign agency’s various requirements.
The picture that emerges here – that in the event of a possible imminent Russian armed attack on a country in Sweden’s immediate vicinity, the country’s political leadership would have been for the most part incapable of seeing beyond MUST’s intelligence assessments – ignores the fact that the government also receives a wealth of foreign intelligence from its embassies in foreign capitals, and political input through other channels – the staffs of politically appointed officials under certain ministries, the Riksdag, party organizations, and so on. The inquiry did note that the Foreign Ministry and the embassies “contribute … extensive reporting on developments in the world around us,” but nevertheless chose “not to examine these activities in detail” (SOU 2025:78, p. 63).
It seems surprising that the Inquiry paid no attention to sources such as reports from the Swedish embassy in Moscow, or from other Swedish embassies of relevance here. Had it done so, the question of parallel structures, a civilian intelligence agency and a foreign ministry engaged with largely the same kind of foreign intelligence, which Hirschfeldt cited as a reason against establishing such an agency, would have been hard to avoid.
The Inquiry also failed to address the role of the media with regard to the flow of information to individual ministers and their respective ministries. This also seems unexpected in light of the comprehensive reporting in both Swedish and international media in late 2021 and 2022, primarily on U.S. intelligence estimates according to which a Russian attack on Ukraine was soon to expected. The Inquiry did note that throughout the autumn of 2021 both the United Kingdom and the United States published intelligence assessments where “a full-scale invasion of Ukraine [was held to be] the most likely scenario” (SOU 2025:78, pp. 78, 86). It failed, however, to address how these assessments might have informed the Government’s understanding of the international situation. The Inquiry did note that MUST’s assessment “had a significant impact” on the Government’s view of events (SOU 2025:78, p. 86); such a formulation does not rule out the possibility that other factors may also have been significant, but since no other factors were mentioned, the picture of MUST’s responsibility for the alleged Swedish intelligence failure is reinforced.
It may be objected here that the Inquiry’s terms of reference simply did not allow for an analysis of the flow of intelligence from other sources, or of the way the government interpreted the intelligence. Even so, and granting that the Inquiry was bound to steer clear of strictly political matters, it still seems questionable to present MUST so unequivocally as responsible for the alleged misjudgement of Russian intentions.
It should also be noted that the Inquiry’s terms of reference contain no instruction that it focus on the run-up to the war in Ukraine or the actions of the Swedish intelligence services in that connection. With regard to the run-up phase, the Inquiry can thus be said to have made a broad, or expansive, interpretation of the directives. Yet with regard to the communication of intelligence to the political level during that same phase, which an equally expansive interpretation would reasonably have included, the Inquiry evidently made a narrow one. It is clear, for example, that the Inquiry “shall assess the intelligence needs that intelligence activities should meet“, and if we are speaking of needs, we are at the political level (SOU 2025:78, p. 253).
Intelligence Failure
We now turn to the nature of the alleged intelligence failure. Judging by the reasoning and conclusions of the Intelligence Inquiry, this failure concerns the fact that MUST’s incomplete warning “had a significant impact” on the government’s assessment. While the fact that the government was – possibly – given misleading information can be seen as problematic, a more detailed assessment of the damage caused to Sweden would have been required for an intelligence failure to be unequivocally identified. Has Sweden suffered any damage as a result of MUST’s possible misjudgement of Russian intentions? Did the government suffer any damage at the time? Can we see repercussions today, four years after the outbreak of war?
It should be noted at the outset that MUST has sought to nuance the picture of failure. In an interview for Aftonbladet on 18 June 2025, an unnamed chief of staff at MUST, referring to a special statement sent by MUST to the Inquiry on 3 June 2025, said that “the picture is more nuanced than Bildt makes it seem” (Svensson, 2025). In an interview in SVT’s news documentary Processen, broadcast on 6 July 2023, the then-newly appointed head of MUST, Thomas Nilsson, said that he did not “share the picture” that MUST had failed to raise the alarm about an impending Russian invasion and that “we made assessments that I still believe are valid” (Granlund, 2023b). In an interview on 25 June 2025, Nilsson, in response to the question of whether the whole thing should be described as a failure, said that “I would probably say that it was part of such a failure” and that “the important thing for me is that we have learned from it” (Tidningarnas Telegrambyrå, 2025).
By way of comparison, it may also be noted that the term “intelligence failure” appears never to have been used officially in relation to the 2017 terrorist attack on Drottninggatan in Stockholm, carried out by Rakhmat Akilov, or the espionage directed at the Security Police over several years by Peyman Kia, who was subsequently sentenced to life imprisonment. Yet the damage in these cases was very significant. In the Akilov case, five people were killed; in the Kia case, strategically important information was passed to Russian military intelligence in quantities considered to be on a par with those of Stig Wennerström and Stig Bergling (Granlund, 2023a). Apart from an apparently rather shallow inquiry into the Akilov attack, whose report remains classified to this day, carried out by Runar Viksten, a judge of the Foreign Intelligence Court, neither case has been subject to the kind of official post-mortem seen in the Intelligence Inquiry’s (itself insufficient) investigation into MUST’s alleged failure.
As the Inquiry’s report noted, there is “a relatively rich literature and body of research” on intelligence failures. This consists primarily of detailed case studies of one or more armed attacks or strategic surprises, such as the Japanese attack on Pearl Harbor in December 1941 (Wohlstetter, 1962), the Soviet deployment of nuclear missiles in Cuba in the late summer and early autumn of 1962 (Knorr, 1964), the Arab coalition’s attack on Israel in October 1973 (Bar-Joseph, 2005), and the terrorist attack on the World Trade Center in September 2001 (Parker & Stern, 2005; Zegart, 2007). There are also studies of a more purely theoretical nature (Bar-Joseph & McDermott, 2017; Betts, 1978; Gentry & Gordon, 2019; Jones & Silberzahn, 2013; Levite, 1987; Lillbacka, 2022; Wirtz, 2024).
Common to both the case studies and the more theoretically oriented works is that the damage resulting from an intelligence failure, or a “strategic surprise” (Levite, 1987, p. 1), the latter being something of an equivalent concept, is described in more or less detail. This dimension of damage also features in the definitions of the phenomenon found in the scholarly literature. Levite defines “strategic surprise” as “the sudden realization that one has been operating on the basis of an erroneous threat assessment that results in a failure to anticipate a grave threat to “vital national interests” (1987, p. 1). Lillbacka (2022, p. 27) refers to “errors in collection and/or analysis, and/or errors in decision based on intelligence, having identifiable detrimental consequences in relation to policy aims”.
Although material damage in terms of destroyed warships or buildings is not explicitly mentioned in the definitions, there is nevertheless talk of fairly concrete damaging effects – according to the first definition, in relation to particularly important national interests that, due to an incomplete understanding of the threat, are now unprotected from hostile influence and/or destruction, and in the second definition, “concrete consequences of a harmful nature” in relation to political objectives. The extent to which MUST’s purported inability to foresee the Russian invasion corresponds to the scientific definition of the concept of intelligence failure is debatable. It could be argued that the first part of the first definition is satisfied: Sweden appears to have based its threat assessment on an incorrect analysis of Russian intentions; but not necessarily the latter part, which concerns the inability to foresee a serious threat to core national interests.
Sweden’s capacity to anticipate and avert Russian military action directed against its territory, or against important interests in the Baltic Sea region such as the territorial integrity of other EU member states, has long been one of the country’s key national interests. The risk of such action being directed against any of these interests at the time of the Russian advance on Ukraine was, however, then considered limited: Russian military resources were concentrated on the border with Ukraine, and it was understood that such action, directed against the Baltic states, for example, would have drawn Russia into an armed conflict with NATO (Haglund, 2022; Olsson, 2022). If it was not so immediately before the Russo-Ukrainian war, since 2023 Ukraine’s continued existence as an independent state has been elevated to a conspicuously important goal of Swedish foreign policy. In the foreign policy declaration for 2024, for example, Foreign Minister Tobias Billström noted that “support for Ukraine [is] the government’s foremost foreign policy task in the coming years” (Sveriges Riksdag, 2024a), and in the corresponding declaration for 2025 his successor, Maria Malmer Stenergard, announced that “support for Ukraine is the government’s foremost foreign policy task” (Sveriges Riksdag, 2025a). Particular mention should be made of the then-foreign minister, Ann Linde, who, in her foreign policy statement presented in the Riksdag chamber on 16 February 2022, referred verbatim to the main content of Sweden’s 2009 so-called solidarity declaration, possibly (even probably) to make clear that Swedish solidarity in a military sense ended at the EU’s eastern border (Sveriges Riksdag, 2022a).9
But why is it important to specify the nature of the damage? Is it not enough, as the Inquiry did, to say that “MUST’s ongoing intelligence assessments had a significant impact on how the government assessed developments”? (SOU 2025:78, p. 86). The problem, however, is that if we, like the Inquiry, start talking in terms of a causal chain that includes assumedly inadequate information from MUST as the “independent variable” and the Government’s assessment as the “dependent variable”, then questions about the extent and nature of the damage to this assessment, as well as whether we can identify other sources of information available to the Government at the time, can hardly be overlooked.
This, in turn, brings us back to the political level.
The alleged intelligence failure involves the bureaucratic and the political levels, regardless of whether the Inquiry’s terms of reference allow for such an analysis or not. If, as Bildt has repeatedly emphasized, we are to learn lessons for the future and clarify how such lessons should be acted upon to the benefit of state institutions, it seems equally difficult to ignore such a discussion and analysis. When it comes to institutional lessons, however, the Inquiry has a clear answer: the solution consists, at least in part, of a redistribution of tasks within the Swedish intelligence service and the establishment of a new civilian foreign intelligence agency.
The next-to complete silence on the part of the Inquiry regarding the path from problem identification (intelligence failure) to solution (the establishment of a new authority, the creation of a national intelligence chief, etc.) appears difficult to defend, even with reference to the limitations that the terms of reference can be said to entail. It can, of course, be argued that the Inquiry dealt with such security-sensitive activities that a full analysis, either of possible dysfunctions in terms of analysis and communication at MUST, or in terms of the government’s assessment, was not possible. In a June 2025 interview with Aftonbladet in connection with the presentation of the Inquiry, Bildt noted:
The wording regarding the failure of Russia’s invasion has been carefully considered and agreed upon. I understand that this is a sensitive issue. I would have liked to say more, as a lot of work went into this, but they were a little overly concerned about confidentiality. (Svensson, 2025).
On the other hand, it can be pointed out that, according to the same article in Aftonbladet, MUST provided a statement to the Inquiry in which it related “that it had conducted a series of internal investigations … but that perspectives and accounts of this are not included in Bildt’s inquiry” (Svensson, 2025). This suggests that MUST would have liked to see a more in-depth discussion of these aspects in the Inquiry’s report and that confidentiality did not – possibly – pose an obstacle to the extent that Bildt claimed.
In summary, the Inquiry provided no guidance on the question of harmful effects. Holmström, by contrast, addressed events at the political level, in addition to the bureaucratic level, more specifically two such events: first, the exchange between Blinken and Linde in December 2021; and second, Andersson’s dissatisfaction with the documentation provided by the Ministry for Foreign Affairs and the Ministry of Defence Holmström, 2022). The first case concerns a possible diplomatic failure that may have been caused by misleading information from MUST. Although the information about the content of the conversation is limited to a few lines in Holmström’s article, since the record of the meeting has not been published, it seems questionable in retrospect to have presented information that “spoke against an invasion” to the foreign minister of another country, especially one whose foreign intelligence resources far exceed Sweden’s.
To name the source of that information, the Swedish intelligence service, likewise seems questionable. We know no details of which reports from the Swedish intelligence service the foreign minister relied on; we have no knowledge of the wording or conclusions of those reports; we do not know whether the foreign minister’s interpretation of them was reasonable given their content; and we do not know, as noted, how the words were phrased during the conversation with Blinken.10 But this remains, nonetheless, the closest we come to a description of possible damage to Sweden, the government and the Swedish intelligence services.
The second adverse effect concerns trust within the government, and between different parts of the Government Offices, over their respective assessments. Here we can hardly speak of damage to Sweden in any broader sense, but a different picture emerges, with partly new actors, the Ministry of Defence and the Ministry for Foreign Affairs, said to have provided misleading information to the prime minister. It seems possible that parts of this information rested on the overall intelligence assessment, with MUST as the source; but it seems equally possible, even probable, as noted above, that the documentation contained information from other sources, and that this was supplemented, in the course of the Government Offices’ handling of it, with assessments and conclusions of a more political nature.
Jonsson (2024) notes in his analysis of Swedish intelligence in the run-up to the Russio-Ukrainian war that there may have been a tendency among “key intelligence consumers” to place their trust in the less alarming reports, “particularly as the alternative suggested that Sweden might have to consider NATO membership – a very divisive course of action within the governing Social Democrats, and one that both the Ministers of Defence and Foreign Affairs objected to” (Jonsson, 2024, p. 450). Jonsson does not go into detail on the political dimension of the purported intelligence failure; his analysis is mainly focused on the issue of possible dysfunctions within MUST in terms of groupthink, mirroring, and the like. But it should be noted here that two leading representatives of the government – the Minister of Defence and the Minister for Foreign Affairs – made statements in the months prior to the invasion that can be said to have bound the government to a certain interpretation of Russian intentions. At the Social Democratic Party congress in November 2021, Defence Minister Peter Hultqvist noted that “the others know very well where we stand, and there will be no applications for NATO membership as long as we have a Social Democratic government. As long as I am defence minister, I will definitely never participate in such a process” (Dagens industri TV, n.d.).
On 24 January 2022, at a time when the United States, among others, had decided to donate military equipment to Ukraine, Linde announced that Sweden did not intend to support Ukraine through donations of such equipment. Linde noted that “we have very strict rules against arms exports, so this is not an option for us. But we support Ukraine in many other ways, above all through very high levels of aid for reforms in Ukraine, which we believe is one of the most important things” (Wicklén, 2022). On 13 February 2022, in an interview with SVT Agenda, when asked about her opinion on the ongoing deliveries of military equipment from the United Kingdom and Lithuania to Ukraine, Linde further noted that “the fact is that each country must decide for itself how to support Ukraine. In Sweden, we have one of the world’s strictest export regulations when it comes to weapons, and there is an authority that monitors this, but I cannot imagine that this is a possibility for Sweden.” When asked why it would not be possible, Linde clarified:
The fact is that we have one of the strictest arms export control regulations, and it must be observed even in cases such as this, or perhaps especially in cases such as this. At the same time, Sweden is one of the largest donors to Ukraine’s resilience, with over 240 million per year, and we are also participating in Canada’s Unifier mission to strengthen Ukraine’s armed forces. So we are one of the countries that supports Ukrainian resilience and Ukrainian civil society the most. … We cannot suddenly say that just because there is now a conflict, we should have our arms export regulations revised. We have [such regulations] precisely because they must be observed also when there is a conflict. (SVT Agenda, 2022)
During her presentation of the annual declaration on Sweden’s foreign policy in the Riksdag as late as 16 February 2022, a little more than week before the Russian invasion, Linde again referred to “the strict export control regulations we have” as an obstacle to Swedish support for Ukraine in the form of military equipment. She also referred to an altogether new factor – the Swedish Armed Forces:
I would also like to say that the Swedish Armed Forces naturally have views on this, as we are currently in the process of upgrading our own defence capabilities. That is why I believe that the best course of action for us is to continue our exceptionally high level of support for Ukraine. Among other things, we support the defence forces through training programs. We are participating in Operation Unifier, in the SMM and in other ways to strengthen the defence sector, as well as by providing resilience support. This is the best course of action for us, rather than getting involved in this now. That is how we see the matter (Sveriges Riksdag, 2022a).
It should also be noted here that Linde emphasized in the same foreign policy declaration that “the government [did not intend] to apply for NATO membership”, that “the security policy line remains unchanged” and that “our military non-alignment serves us well and has contributed to stability and security in northern Europe” (Sveriges Riksdag, 2022a). To these statements can also be added Prime Minister Andersson’s assertion on 8 March 2022 that a Swedish application for membership “in this situation … would further destabilize the situation in Europe” (Carlén, 2022).
The Politicization of Intelligence
The statements by Andersson, Hultqvist, and Linde lead us to a third objection to the conclusions of the Bildt Inquiry: its lack of analysis and reasoning regarding, first, how the intelligence from MUST was interpreted at the political level, and second, what the flow of information and impulses in the opposite direction, from the political level to MUST, may have looked like. Andersson’s statement, admittedly made around two weeks after the Russian attack but telling nonetheless, together with the statements by Hultqvist and Linde discussed above, indicates a foreign and security policy language at the political level that left little or no room for a scenario in which Russia would move from words to action and launch a large-scale invasion of Ukraine. Were that to happen, it would be difficult, if not impossible, as indeed it proved, to maintain Sweden’s non-alignment and its policy of not supporting Ukraine with military equipment. As for the flow of information and any political impulses from the Government Offices to MUST, the Inquiry claims to have “reviewed written reports from the intelligence services to the government and the Government Offices over a longer period” and “held discussions with relevant parts of the intelligence system and officials in the Government Offices” (SOU 2025:78, pp. 22–23).11
Although this description of how the investigative work was structured leaves open the possibility that political signals to the intelligence community were discussed, no such reasoning appears in the Inquiry’s report. It is difficult, if not impossible, to retrospectively reconstruct even the most basic elements of the communication between MUST and the Government Offices from the outside. Here I am concerned specifically with the briefings given to representatives of the Government Offices, or to ministers, at which signals of a political nature may have been conveyed, rather than with the frequency or content of MUST’s reporting (which is in any case likely to remain classified for the foreseeable future).
As for the Riksdag, the task is simpler, at least in terms of mapping whether and when communication has occurred. Here it appears that on 9 December 2021, the head of MUST, Hallin, together with representatives of FOI and the Swedish Institute for International Affairs, informed the Defence and Foreign Affairs Committees “about the security policy situation in Ukraine” and that “the deputies’ questions had been answered”. Thus, at least in theory, both members of the coalition parties and parties in opposition were given the opportunity to condition what was acceptable to say and not to say (Sveriges Riksdag, 2021a, 2021b). There is no information on further presentations to the Riksdag during the period up to the outbreak of the war.
Overall, this implies that the Inquiry’s claims of an intelligence failure deriving from MUST’s analytical, organizational and communicative shortcomings should be regarded as poorly substantiated. In part, this is because the analysis of these shortcomings is incomplete – as mentioned, the Inquiry has failed to address the two questions, firstly, why MUST’s analysis was not adapted to the information that the Inquiry believes “pointed in different directions” – and, secondly, in what way(s) the purported organizational dysfunctions with MUST affected the quality of its analytical work. In part, because the Inquiry failed to address the political level and its set of political blockages, as well as the flows of information and direction from this level to MUST. The conclusion can thus hardly be anything other than that, in attempts to explain the alleged Swedish intelligence failure, references to the political level must be regarded as no less valid than references to MUST. It likewise seems justified to conclude, in the light of the statements by Andersson, Hultqvist, and Linde discussed above, that what we are dealing with here is primarily a failure at the political level. This took either of two forms, and possibly both: politically conditioned “blinders” among certain key circles within the Government Offices in their reading of MUST’s assessments, or MUST’s assessments being shaped by more or less pronounced signals of a political nature from the government – that is, the politicization of MUST. If the latter, it may have occurred during the phase preceding the outbreak of war, or, possibly, already before it. Give that the detailed content of the reporting remains inaccessible, it is not possible to make any final judgement here.
These pathologies – that decision-makers see what they want to see and that the intelligence services consciously or unconsciously adapt their assessments and their language to what they believe to be in line with political goals and objectives – are well-known from research on the so-called politicization of intelligence activities (Bar-Joseph, 2013; Coletta, 2018; Kent, 1949, pp. 195–201). Interestingly, the Inquiry raises the issue of the risk of politicization, albeit in the context of its reasoning about MUST’s “circumstantial and unclear” presentations and written reporting mentioned above.
The Inquiry noted here that “a development work … in [the Government Offices] is underway to increase customer benefit and ensure that intelligence offers greater relevant support in central policy processes” (SOU 2025:78, p. 146). In this context, the Inquiry further noted that “the dialogue with the intelligence agencies needs to be organized in a way that counteracts the risk of intelligence assessments being adapted to what is believed the recipients want to hear or benefit from, at the expense of factual accuracy,” and that “even information that recipients may perceive as going against desirable policy needs to be presented” (SOU 2025:78, p. 146). Regrettably – or rather, remarkably – the Inquiry failed to reflect on the possibility that this is exactly what we see in the dialogue between MUST and the Government Offices in the run-up to the Russo-Ukrainian War.
Conclusion
If the answer to Sweden’s alleged intelligence failure should be sought at the political level rather than with MUST, as has been argued above, one of the Inquiry’s most central arguments for establishing a new foreign intelligence service falls apart. However, the Inquiry also cites, as mentioned above, several other reasons why such an authority, with an additional series of further reforms and legislative amendments, is necessary. Some of these can probably be said to be well-considered; others, less convincing. And it would be incorrect to claim that all change is necessarily bad in this situation.
As mentioned above, the Intelligence Inquiry highlights the importance of learning from intelligence failures (SOU 2025:78, p. 23). The lessons that should correctly be drawn are, however, far from as obvious as the Intelligence Inquiry would like to suggest.
And here, there are reasons to recall one of the more insightful analyses of intelligence failures – that of the American political scientist Richard K. Betts.
“In the best-known cases of intelligence failure,” Betts points out in his 1978 article “Analysis, War, and Decision: Why Intelligence Failures Are Inevitable”, “the most crucial mistakes have seldom been made by collectors of raw information, occasionally by professionals who produce finished analyses, but most often by the decision makers who consume the products of intelligence services” (Betts, 1978, p. 61).
Observers who see notorious intelligence failures as egregious often infer that disasters can be avoided by perfecting norms and procedures for analysis and argumentation. This belief is illusory. Intelligence can be improved marginally, but not radically, by altering the analytic system. The illusion is also dangerous if it abets overconfidence that systemic reforms will increase the predictability of threats. The use of intelligence depends less on the bureaucracy than on the intellects and inclinations of the authorities above it (Betts, 1978, p. 61).
Notes
[1] For an overview of the Swedish donations of military equipment, see Government Offices of Sweden (2025a). See also the government’s written communication to the Riksdag (Skr. 2025/26:162, 2026). On the decision-making processes and parliamentary oversight concerning the Ukraine donations, see Matz (2025).
[2] Government bills are referred to as proposition (prop.); parliamentary committee reports are referred to as betänkande (bet.).
[3] For an overview of Sweden’s intelligence, see Agrell (2013).
[7] The Sandler Inquiry was headed by the Swedish former Foreign Minister, Rikard Sandler, it was in operation in 1945–1948, and produced aside from the fourth report, three major reports (SOU 1946:36, SOU 1946:93, and SOU 1948:7). The Tammelin Inquiry, headed by the Judge with the Stockholm City Court, Erik Tammelin, was in operation 1947–1948.
[8] The 1996 Intelligence Committee proposed that the Government Offices be provided with a structure for this purpose – a coordination function for intelligence matters consisting of an interdepartmental preparatory group at the state secretary level supported by a coordination secretariat for intelligence matters. The central task will thus be to ensure that the government continuously has access to the combined contributions of the Swedish intelligence service to a consolidated security policy analysis of acute or long-term threats to or risks for the country’s external security (SOU, 1999:37, pp. 34–35).
[9] Linde said here that “Sweden will not remain passive if a disaster or an attack were to affect another EU country or a Nordic country. We expect these countries to act in the same way if Sweden is affected. An armed attack against Sweden cannot be ruled out. Therefore, we must be able to give and receive support, both civilian and military” (Riksdagens protokoll, 2022). For the solidarity declaration, see prop. 2008/09:137, and Riksdag (2009).
[10] In an interview with Holmström on September 8, 2023, Linde confirmed that she had disputed the American intelligence reports about an imminent Russian attack with reference to an assessment “that had been made together with virtually all European countries”. However, she did not provide any further details about the conversation (Holmström, 2023).
[11] The work process regarding the analysis of the intelligence failure is also described elsewhere in the report, where it states that the Inquiry “has particularly studied the intelligence reporting in connection with the Russian buildup against Ukraine during 2021 and the beginning of 2022,” and that it has conducted “a number of interviews with individuals in central functions during this period” (SOU 2025:78, pp. 63–64).