Justice Department Inspector General Michael Horowitz’s long-awaited report on the FBI’s “Crossfire Hurricane” investigation is finally out, and notwithstanding furious efforts from all quarters to claim otherwise, it fails to neatly validate anyone’s favored political narrative.
Contra the hopes of Donald Trump’s more ardent admirers, it fails to turn up anything resembling a “Deep State” cabal within the FBI plotting against the president, or deliberate abuse of surveillance authorities for political ends. Yet it also paints a bleak picture of the bureau’s vaunted vetting process for warrant applications under the Foreign Intelligence Surveillance Act, documenting a host of material omissions or misrepresentations in the government’s case for wiretapping erstwhile Trump campaign adviser Carter Page, whose privacy was invaded for nearly a year on disturbingly thin grounds. Though it does not describe an investigation motivated by political bias, it is a textbook account of confirmation bias that should raise disturbing questions about the adequacy of the FISA process—and not just in this investigation.
The heart of the Horowitz report deals with the Carter Page FISA application, and documents a progression that ought to sound familiar to anyone who’s studied the history of the intelligence community: An investigation begins with a kernel of reasonable suspicion, and facts are marshaled to support a theory. As it gathers momentum, those initial suspicions congeal into assumptions. New information that fits the original theory is added to pile of evidence—while a growing body of contradictory of information is overlooked. It’s possible to read the Horowitz report and think that the initial 90-day wiretap of Page was justified, but far harder to rationalize intrusive surveillance that carried on for nearly a year, through three separate renewals, even as evidence mounted that should have undermined the basis for the warrant.
As the report recounts, “Crossfire Hurricane”—the FBI’s codename for its probe of potential links between Russia’s election interference operation and the 2016 Trump presidential campaign—originated in the summer of 2016 with a tip from the Australian government (a “friendly foreign government” in the report): Trump campaign aide George Papadopoulos had been drunkenly repeating an academic acquaintance’s startling assertion that the Russian government had thousands of potentially damaging emails related to Democratic candidate Hillary Clinton. The intelligence community was already seeing the outlines of an unprecedentedly brazen, multi-pronged effort to meddle in the presidential election. Now here was an indication that the Trump campaign might be not just an unwitting beneficiary of Russian efforts, but a knowing participant. The FBI quickly focused on four individuals in Trump’s orbit with ties to Russia: Page, Papadopoulos, campaign chair Paul Manafort, and national security adviser Michael Flynn.
While FISA surveillance of Page and Papadopoulos was apparently contemplated in August, Justice Department attorneys determined investigators lacked probable cause to establish that either was acting as an “agent of a foreign power,” the critical showing they’d need to make to the Foreign Intelligence Surveillance Court. That changed in September 2016, when the FBI got wind of former British intelligence officer Christopher Steele’s research into Trump’s Russian ties—opposition research indirectly commissioned by the Democratic National Committee, and now notorious under the collective moniker “The Steele Dossier.” Steele’s reporting, gathered from a network of sources and sub-sources, purported that Page was a key figure in a “well developed conspiracy of cooperation” between the Trump campaign and the Russian government. The FBI would lean heavily on Steele’s reporting in its petition to the FISC for a warrant authorizing electronic surveillance of Page.
As the Horowitz report documents, even that first application, submitted in October 2016, contained a series of notable omissions or misstatements, though whether they would have made a difference to the FISC’s assessment of the evidence is anyone’s guess. Among the most significant:
• The application failed to mention Page’s relationship with the CIA (“another government agency” in the report), which had designated him an “operational contact,” and the fact that Page had provided the Agency with information about his previous contacts with Russian intelligence officers—contacts that were part of the basis for suspecting Page had been recruited to act as an “agent of a foreign power.”
• It described Steele as a source whose previous reporting had been “corroborated and used in criminal proceedings,” which overstated Steele’s role in the Justice Department’s investigation of corruption in the international football league, FIFA. Though Steele had apparently provided information critical to launching the FBI’s FIFA investigation—leading to a raft of indictments and guilty pleas in 2015—Steele’s reporting had not actually been introduced as evidence in court. Since the Page application leaned heavily on Steele’s reporting, which it could not independently confirm, the precise characterization of his credibility as a source was important to informing the FISC’s assessment of how much weight to afford his claims.
• Both Page and Papadopoulos had made statements to FBI informants denying various contacts alleged by Steele’s reports. Page had specifically denied taking part in a meeting with a pair of Russian oligarchs described by Steele, and claimed to have little contact with Manafort, though Steele’s account had Page acting as an intermediary between Manafort and the Russian government.
• In an interview with the FBI, Steele had characterized a particular individual presumably Sergei Millian, though he is not named in the report—as a “boaster” given to “embellishment.” Though FBI analysts had independently identified this individual as the likely source of key claims about Page in Steele’s reporting, this characterization was not included in the application.
• The FBI assessed that Steele had not directly provided information about his research to a reporter for Yahoo News—an assessment that would ultimately prove incorrect—but nothing in the case file provided a documentary basis for or explanation of that assessment.
These omissions matter because FISA applications typically remain secret forever—indeed, the Page application is the first to ever become public even in part. That means not only does the FISC rely on the government to present it with a complete picture, including facts that might call the reliability of government sources into question (something that’s true of every wiretap application) but there’s typically little risk that an agent who submits a tendentious affidavit supported by cherry-picked evidence will have to defend their work in the harsh light of an adversarial proceeding, such as a criminal trial. If the government isn’t forthright about presenting evidence that cuts against a finding of “probable cause,” as well as the evidence for it, they’re unlikely to be held to account.
It’s worth emphasizing, though, that the picture Horowitz paints remains fundamentally at odds with claims that FBI or DOJ leadership conspired to mislead the FISC, plotting to use surveillance of a peripheral campaign adviser as the linchpin of some Rube Goldberg–ian scheme to undermine the Trump administration. Each of these gaps represents information that lower-level case agents failed to recognize as material to the application—and in at least one case, an agent providing an erroneous response when a DOJ attorney asked for clarification about precisely when Page had been a source of information for CIA.
What they do show, however, is that the much-ballyhooed Woods Procedures, designed to ensure that representations to the FISC match the information in the FBI’s case files, are no guarantee that the Court is getting a complete picture. Woods review will catch a claim that’s unsupported, but it won’t reliably ferret out information weakening the government’s case that’s not flagged as relevant by the agents working the case. (The report does identify a handful of facts that passed through Woods review without apparent documentary support, though the instances in the initial application, at least, are trivial.)
One exception, perhaps ironically, concerns the charge that the FBI had deceived the court about the politically motivated funders of Steele’s research: A lengthy footnote discussing just that was added to the application at the urging of DOJ attorneys, though in keeping with the general practice in FISA applications, specific American people and groups (such as “Donald Trump” or “the Democratic National Committee”) were not mentioned by name.
It’s with the renewal applications, however, that the omissions start to get truly egregious:
• By January 2017, FBI had identified and independently interviewed Steele’s “primary sub-source,” the results of which should have provoked a serious reevaluation of the reliability of Steele’s reporting. The source gave accounts that conflicted with Steele’s reports in certain respects, or characterized information he’d passed on as little more than rumor and thirdhand gossip. The renewal application did mention that FBI had spoken with the source, but didn’t draw attention to points of contradiction, or the source’s own skepticism about the information he’d passed on to Steele.
• An FBI attorney added a line to an email from a CIA liaison saying that Page had not been a source—when the opposite was true.
• Though FBI had obtained evaluations of Steele’s reliability from former colleagues, some of which questioned his judgment, the renewal applications recapitulated the initial assessment of his credibility unaltered. Nor did they include a qualification from a formal FBI source validation report that, while Steele had provided some valuable information in the past, much of it had never been definitively corroborated.
• The renewal applications never corrected the initial assessment that Steele had not provided information to Yahoo News, though FBI officials had learned this was wrong before the final renewal.
• The renewals continued to include assertions that Page had played a role in modifying the GOP platform to weaken language committing to aid Ukraine against Russia, and made no mention of mounting evidence against Page’s involvement.
• Acquisition of Page’s older electronic communications revealed minimal contact with Manafort, but FBI failed to note this, or otherwise qualify its reliance on Steele’s characterization of Page as Manafort’s liaison to Russia.
Horowitz also faults the FBI for some omissions that seem less problematic: DOJ official Bruce Ohr had characterized Steele as “desperate” to prevent Trump’s election, which if Steele believed his own reporting seems entirely natural, not suggestive of preexisting bias. And though the bureau did eventually learn the DNC was the ultimate source of funding for Steele’s research, it did not add this to its original footnote indicating a likely political motive for the assignment—a detail that seems both immaterial and redundant: Any judge with the minimal wit needed to identify the unnamed candidate referenced in the application could presumably also deduce the identity of his political opponents, if it mattered.
Nevertheless, the gaps and inaccuracies in the renewal applications are clearly more serious than those in the first. It is not hard to imagine the first warrant being granted even if FBI had included in its application all the information available to it at the time; it is almost impossible to believe the same of the last one. Yet it remains, if anything, still more incoherent to suppose that, in June 2017, the FBI was seeking to prolong increasingly fruitless surveillance of a former campaign adviser—surveillance long since known to the target—as part of an attempt to somehow harm Trump.
Instead, the picture that emerges from the Horowitz report is not so much sinister as banal: The government asked the court for “one more go” essentially out of inertia. Case agents weren’t motivated to think terribly hard about whether the most recent piece of information they’d uncovered contradicted a claim they’d made to the court months earlier. Higher-level attorneys reviewing renewal applications focused almost exclusively on vetting the new information in each filing, never going back to reexamine earlier assertions and test whether they were still defensible. Verification meant checking the files to validate fresh data, but not to take a fresh look at early assumptions. After all, someone had checked that already, right?
This picture is, in its own way, and for very different reasons, as disturbing as the image of a “Deep State” cabal with a vendetta against Trump: Vendettas are at least specific. The grave defects in the surveillance of Page seem more likely to be symptoms of a more apolitical, and therefore more systemic, form of bias. Their underlying causes—reliance on sources whose claims are hard to directly check, imperfect information, case agents making judgments about which facts in a vast sea of data might be legally material—aren’t peculiar to elections but endemic to intelligence.
The investigators working Crossfire Hurricane well understood they were charged with a Sensitive Investigative Matter—one destined to draw a level of scrutiny unprecedented in the history of FISA. Under the circumstances, you might expect them to operate with especially scrupulous exactitude. If the Horowitz report reflects what we find when we start turning over rocks under those conditions, what kind of errors and omissions might we expect to uncover in the case files of FISA targets less likely to inspire congressional hearings? It’s past time to find out.
Support our independent journalism
Readers like you make our work possible. Help us continue to provide the reporting, commentary, and criticism you won’t find anywhere else.