Documented conduct leaves competing Havana syndrome concealment claims unresolved

Summary

The forthcoming book by Michael Rey and Oriana Zill de Granados, with corroboration from Stanford professor David Relman, alleges that parts of the US intelligence community concealed evidence linking anomalous health incidents to a Russian directed-energy program.
Two sources confirmed to The Guardian that US authorities acquired a suspected “Havana syndrome-style device” from a dark-web seller in late 2024 — reportedly man-portable, non-thermal, pulsed-microwave — and brought it to the US for animal testing reportedly showing AHI-similar symptoms.
Tulsi Gabbard, then director of national intelligence, rescinded both the 2022 scientific-panel finding and the 2023 intelligence-community assessment that foreign adversaries were “very unlikely” behind the incidents, citing “selective exclusion of evidence” and “mischaracterisation of sources” without releasing new evidence or a revised assessment.
The substrate documents conduct consistent with former CIA director Bill Burns’s “rigorous effort” framing and with Relman’s “selective exclusion” framing, leaving the competing characterizations unreconciled in the public record.

A forthcoming book by two longtime CBS 60 Minutes producers alleges that Russia was behind at least some cases of anomalous health incidents (AHIs) — the condition popularly known as Havana syndrome — and that parts of the US government and intelligence community concealed supporting evidence, according to The Guardian’s account of the work. The allegation lands in a public record in which two senior figures with overlapping access — Stanford professor David Relman, who chaired the 2022 scientific panel, and former CIA director Bill Burns — offer directly contradicting characterizations of whether evidence was concealed, and in which the most recent formal intelligence assessment was rescinded this summer by then-director of national intelligence Tulsi Gabbard without a replacement judgment being released. The Guardian drew on early access to the book and its own interviews with intelligence sources, AHI patients and scientists. The central analytical question — whether pulsed-radiofrequency exposure caused specific symptomatic clusters — sits in a public record that contains documented conduct consistent with each competing characterization, and the divergence between Relman and Burns marks the analytic frontier the public record has not yet crossed.

What the book alleges

The forthcoming book “The Havana Syndrome,” by Michael Rey and Oriana Zill de Granados — producers at CBS’s 60 Minutes who covered the Obama administration’s 2015 reopening of diplomatic relations with Cuba — argues that Russia was behind at least some episodes and that parts of the US government and intelligence community concealed evidence supporting that conclusion, The Guardian reported. The producers’ investigation began when, after reports surfaced in 2016 of a mysterious condition affecting employees at the reopened US embassy in Havana, Zill contacted a senior US diplomat posted there. “I expected him to say: ‘This is all bullshit,’” Zill told The Guardian. “But he said: ‘Oh no, this is completely true. And I can put you in touch with a bunch of people who were hit.’” The pair spoke to more than a dozen affected people off the record, then began receiving reports of similar cases among US diplomats in China and whispers of cases in other locations around the world.

Once 60 Minutes aired its first episodes on the topic, the producers wrote, AHI patients including undercover CIA operatives whose assignments were classified began reaching out. The pair documented a geographic-occupational pattern, Rey and Zill reported: the alleged victims were working, or had previously worked, on beats where they rubbed against Russian interests. Rey told The Guardian: “I went through years of doubts about the story, and all the crazy things we were hearing… We’re meeting these highly effective intelligence officers and they’re all telling versions of the same story.” The pattern, they reported, held in Cuba — where the original cluster emerged during the Obama-era thaw, which had irritated hardliners in the Cuban government and their backers in Moscow — and in Georgia, Kyrgyzstan, Serbia and Vietnam, allies of Moscow where parts of the elite were looking to improve relations with the West. Several CIA officers who had been stationed in Kyiv around the time of the 2014 Maidan revolution in Ukraine, decried by the Russian president Vladimir Putin as a US-backed coup, later suffered AHIs in different locations around the world.

The competing characterizations

Relman, the Stanford professor who chaired the 2022 scientific panel, told The Guardian that “to any reasonable person looking at all the evidence I was able to see, it inevitably leads to the conclusion that there may have been a cover-up.” Relman’s panel had access to patient records and more than 1,000 classified documents on the topic; he characterized a subset of cases as “most plausibly explained by exposure to pulsed microwave energy” and said the underlying record points to concealment.

Burns, in a comment provided for the book, said it was “simply not true that there was a ‘cover-up’ by the CIA” and described the earlier investigation as “a rigorous effort across the entire intelligence community.”

The public record contains documented acts consistent with each framing. Rey and Zill describe a nascent investigation at CIA headquarters in late 2019 that reportedly found phones linked to Russian intelligence operatives near suspected incidents, including the episodes that year involving the senior CIA officer in London and Warsaw — in London, a suspicious phone was reportedly just 10 metres (30 feet) from his hotel room. When then-CIA director Gina Haspel was briefed, Rey and Zill write, “the meeting abruptly ended after Haspel shouted that if she took reports like this to President Trump, she would be fired as director.” Rey also reported that the FBI sent criminal investigators, rather than counterintelligence investigators with access to classified information, to Havana after the attacks there.

Rey told The Guardian that what the producers characterize as concealment was not the product of a single decision. “There’s not a single cigar-filled backroom meeting that led to decisions getting made,” he said. “There was a lot of, ‘This is a pain in the ass’ or ‘The president’s not going to want to hear this’. There was a bureaucratic dysfunction, careerism and things that just got in the way of proper investigation.”

Gabbard’s rescission

This summer, Tulsi Gabbard, then director of national intelligence, rescinded both the 2022 scientific-panel finding and the 2023 intelligence-community assessment that foreign adversaries were “very unlikely” behind the incidents, citing “selective exclusion of evidence” and “mischaracterisation of sources.” The Guardian reported that the statement released no new evidence and provided no revised assessment.

The 2023 intelligence-community assessment had read: “The available intelligence consistently points against the involvement of US adversaries in causing the reported incidents,” calling a Russian campaign “very unlikely.” A tweaked version issued at the end of the Biden administration said that while five US intelligence agencies still found foreign involvement “very unlikely,” one agency believed there was a “roughly even chance” that a foreign actor had used “a novel weapon or prototype device” in a number of attacks.

Mark Zaid, a lawyer who represents more than 30 AHI patients, welcomed Gabbard’s memo and described it as a first step to overturning what he characterized as a CIA cover-up. When asked why the CIA might have perpetrated such a cover-up, Zaid suggested it could be to protect its reputation after prior negligence. “This has built up to a level that the agency can’t let the truth get out because it’s going to reflect so poorly on them,” he said.

The device

The most significant new piece of evidence to emerge recently, The Guardian reported, is the alleged US government acquisition of a “Havana syndrome-style device” in an undercover operation — first reported this year by the independent journalist Sasha Ingber. Two sources confirmed to The Guardian that a device was acquired and brought to the US in late 2024 for testing after a purchase from a dark-web seller. A person who claimed to be part of the undercover team told Rey and Zill that the device was “man-portable,” shot non-thermal pulsed microwaves and was acquired from “a complex Russian illicit network.” It was not believed to have been used in an actual incident, but testing on animals has reportedly shown it may cause similar symptoms to AHIs, which could prove that building such a device is scientifically possible.

The device’s diagnostic weight runs in one direction. If the device’s animal-testing results are made public and the device’s specifications are independently replicated, the scientific-impossibility objection to the directed-energy hypothesis loses force. A second discriminator would be forensic recovery of physical or electronic signatures at a documented incident location that match the acquired device. A third — and the strongest — discriminator would be HUMINT or SIGINT disclosure, since one source noted that US collection capabilities on Russia are strong enough that absence in known channels is itself diagnostic; the absence of such disclosure is itself part of the record on which the concealment claim rests.

The competing causal hypotheses

The substrate supports five distinguishable accounts of what caused AHIs.

Directed-energy weapon (H1). Relman’s panel concluded in 2022 that “the signs and symptoms of AHIs are genuine and compelling” and “could be due to external stimuli.” Per Relman, scientists who disagreed with H1 were “simply unaware of the research into such technology that had been taking place for decades in several countries, especially in the Soviet Union and then Russia.” A declassified NSA statement included in Mike Beck’s 2014 workplace compensation claim noted that “there is intelligence information from 2012 associating the hostile country to which Mr Beck travelled in the late 1990s with a high-powered microwave system weapon that may have the ability to weaken, intimidate or kill an enemy, over time, and without leaving evidence.” Sources told The Guardian the unnamed hostile country was Russia. Christo Grozev, a journalist and investigator who has exposed several Russian intelligence operations, claimed leaked travel data suggested at least four potential crossovers between suspected AHIs and travel by officers from Unit 29155 of Russia’s GRU military intelligence, known for its campaign of sabotage and assassinations in western countries. Grozev said he had traced travel by “members of what appears to be a technical unit adjacent to 29155” to Cuba during the initial attacks.

The principal discriminator against H1 is collection-channel silence. One source told The Guardian there was no “smoking gun” — no human intelligence or intercept showing Russian operatives discussing a programme to use directed energy against Americans. “Everyone has seen how strong our collection capabilities on Russia are,” the source said, referencing how US intelligence obtained Vladimir Putin’s war plans before his February 2022 invasion of Ukraine. “If this was really a large-scale Russian programme, we would have expected to see something in some of these channels, but we didn’t.” Another former intelligence official disagreed. “If you’re waiting for an intercept of someone saying: ‘My name is Igor and I’m going to press the button on my secret laser’ before you have confidence, that’s probably never going to happen,” the official said. “But for any reasonable person who has access to the intelligence and knows how to read it, the Russian fingerprints are clear.”

Mass psychogenic illness (H2). Sociologist Robert Bartholomew and neurologist Robert Baloh co-authored a book declaring the phenomenon “a case study in the social construction of mass psychogenic illness.” Cellular neuroscientist Douglas Fields argued in a journal article that the medical studies of patients showing cognitive changes had used unsound methodology, and that energy-weapon theories “defy the laws of physics.” A further sub-variant reported in the substrate is the allegation that “some inside the CIA accused those involved of faking their injuries in the hope of a medical payout” — a conscious-fraud-for-gain framing that shares H2’s broader prediction that objective-signs clinical evaluation would be falsifying but treats reports as deliberate fabrication rather than unconscious psychogenic production. The 2021 expansion of reports into the Vienna CIA station, where the work was reportedly paralyzed and the station chief recalled, and onto US soil including the White House grounds, is treated in the substrate as consistent with H2’s prediction that media attention and reporting encouragement produce clusters.

H2’s principal discriminator against H1 is clinical. Relman described features the panel found “near impossibility of explaining… based on known medical and environmental conditions.” James Giordano, professor emeritus of neurology at Georgetown University Medical Center and the forensic neuroscientist on the original cohort of cases from Havana, told The Guardian: “The early scepticism was perfectly appropriate, but over time we started to see an increase in the number of individuals who have gone through clinical evaluations that cannot be faked. We are talking about objective signs that are presenting with a very distinct pattern,” citing “a profound tumbling or falling sensation,” pressure in the head, chest and throat, and cognitive effects. “It was true that chemical or drug exposure could cause similar symptoms,” he said, “but this would be visible on blood tests.”

Environmental or organic cause (H3). Relman conceded that “plenty — perhaps the vast majority — of AHI claims could be explained by other factors.” Substrate-noted environmental mechanisms range from “environmental factors and toxins” in general terms to the specific proposal attributed to unnamed experts that symptoms reflected “the sound of chirping crickets.” The diagnostic weight of the crickets proposal is minimal: it does not predict the geographic-historical breadth of the AHI record and offers no mechanism for the documented clinical findings at Vienna, the London hotel incident, or the Belgrade researcher case.

Data-collection device malfunction (H4). Treated in the substrate as a historical hypothesis — one early theory that AHIs could result from data-collection attempts gone wrong, a machine designed to pull classified data from operatives’ electronic devices with unfortunate side-effects when turned up too high. The substrate reports that the damaging health effects “would have quickly become apparent to the operators,” weakening the hypothesis in its original form.

Mixed causation (H5). Relman’s subset framing — dozens of unusual cases within a much larger population — is consistent with H5, as is the modified end-of-Biden-administration assessment in which five intelligence agencies still found foreign involvement “very unlikely” while one agency believed there was a “roughly even chance” a foreign actor had used “a novel weapon or prototype device” in a number of attacks. The sub-variant — that the smoking-gun absence is treated not as evidence of programme-absence but as evidence of collection-channel prioritization that did not target the relevant operational discussions — is also a form of H5.

Confounders and the identifiability problem

Three confounders are apparent from the recorded record. Selection: officers who worked Russia-related beats were by definition the population most likely to attribute ambiguous symptoms to a Russian attack and to be reported into AHI-tracking systems, generating an apparent geographic-occupational correlation that does not by itself establish causation. Reverse causation: media coverage of AHIs plausibly increased both reporting volume and diagnostic suspicion, producing clusters in time that resemble the temporal signature of an escalating attack campaign. Latent variable: institutional dysfunction at the intelligence community — documented in Rey’s account of an FBI criminal-investigation response rather than a counterintelligence response in Havana, and in Rey’s account of a “nascent investigation” at CIA headquarters in late 2019 that reportedly found phones linked to Russian intelligence operatives near suspected incidents — is consistent both with the concealment characterization and with the dysfunction characterization; the public record does not yet discriminate between them.

The central analytical question — whether pulsed-radiofrequency exposure caused specific symptomatic clusters — sits at the counterfactual rung in the potential-outcomes tradition: would these specific individuals have experienced these symptoms absent weapon deployment? Counterfactual identification requires either an experiment that cannot ethically or practically be conducted, or strong instrumental variables that satisfy the exclusion restriction. The geographic correlation between reported incidents and Russian intelligence operational patterns functions as an instrument-like variable, but fails the exclusion restriction because Russian operational patterns correlate with other features — political relationships, diplomatic posture — that have their own causal paths to AHI reports. The back-door path through “patient awareness of the AHI phenomenon” → symptom interpretation → reported case is not blockable from observational data; conditioning on media coverage to block this path would partially condition on a collider and adjust away the very effect under measurement.

The historical record

The origins of the syndrome go back long before Cuba. Mike Beck, a counterintelligence officer at the NSA, became ill while on a temporary assignment in the 1990s to check a US diplomatic facility for bugs, developed a rare form of Parkinson’s disease at 45, and died this year, The Guardian reported. The first known AHIs in the contemporary batch occurred in late 2014, when several US government employees at the consulate in Frankfurt, Germany, became ill. One was found unconscious in a consulate stairwell after feeling sudden, intense pressure on her chest and hearing a “high-pitched squeal,” Rey and Zill reported. The following year, a researcher at a US defense institute who asked The Guardian not to use her name was on a tour of the Balkans, attending meetings with military officers from the host nations alongside US defense attaches; her itinerary took in Montenegro, which was in the process of applying for Nato membership, and Serbia, a Russian ally that harbored resentments towards the US over the 1999 Nato bombing of Yugoslavia. On her final evening in Belgrade, as she walked through lively pedestrianized streets to her hotel, she felt “a sinking feeling, as though I was in a lift that was descending suddenly.” Inside her hotel room, she continued to feel “a falling sensation, a feeling of gravity pulling at me, and nausea,” and her government laptop began flickering with blue static; it never worked again. In the weeks that followed, she developed pain that felt “like a drill was going into my eye,” struggled to speak clearly and to focus at work, and said she used up her vacation, her sick leave and leave donated by colleagues before eventually leaving her job.

On one reading, this breadth strengthens H1’s patient-pattern correlation: the geographic-historical footprint correlates with Russian intelligence operational reach in the same period. The same breadth cuts against single-cause attributions in either direction: a 1990s-to-present envelope exceeds the working lifespan of any single device-programme hypothesis and resists parsimonious explanation under any single candidate cause. The breadth satisfies the “no immediate or ultimate test” criterion that distinguishes what Rittel and Webber termed wicked problems: each new incident arrives with its own institutional history, classification posture, and patient-cohort politics, defeating replication as a falsification lever.

The information-flow gap

The pattern — that substrate events follow an intelligence-acquisition-to-public-disclosure gap — is itself a documented feature of the AHI record. The late-2024 device acquisition was completed in advance of the summer 2026 Gabbard rescission but its public significance was first reported by the independent journalist Sasha Ingber this year and surfaced as book evidence at 2026 publication. The 2023 intelligence-community assessment finding foreign involvement “very unlikely” was issued without apparent reference to the late-2024 device evidence. The 2014 workplace-compensation statement in the Beck case entered the public record via the workplace-claim process rather than via intelligence-community disclosure.

The gap is analytically distinct from the causal-mechanism question and the institutional-accountability question; it could be reduced or eliminated without resolving either, and its persistence does not by itself adjudicate between H1 and H2 or between the rigorous-effort and selective-exclusion framings.

How this is being framed

Four framings are operative in the substrate. The forensic-attribution framing — dominant in the 2023 intelligence-community assessment — treats the question as a forensic-attribution problem: given a documented population of symptoms, identify the causal mechanism. It concentrates resources on the highest-information question and presumes the symptoms have a unitary cause; Relman’s description of “dozens” of unusual cases within a much larger population supports a heterogeneous-cause framing instead. The stakeholder-position framing varies sharply by who is asking: AHI patients ask whether the harm they experienced will be officially recognized and the perpetrator named; intelligence-agency leadership asks whether the agency’s response has been competent; US-Russia bilateral relations would be affected by an official attribution’s impact on diplomatic posture; relevant Russian institutions (GRU Unit 29155, SVR, FSB) face what disclosure would cost. The historical-genealogy framing locates Havana syndrome within a longer history of suspected directed-energy exposure of US personnel abroad, including the Beck case; the counter-framing holds that earlier reports were not validated and may themselves reflect the same attribution dynamics under examination today. The cross-domain analogical framing treats Havana syndrome as analogous to Cold War allegations of novel Soviet weapons and to the post-1991 disputes over Gulf War syndrome — suggestive but unable to substitute for case-specific evidence.

What happens next

The institutional arc — 2017 first contemporary cases acknowledged, 2021–2023 panel and assessment processes, 2021 expansion of reports across the Vienna station and onto US soil, late-2024 device acquisition, 2026 Gabbard rescission and book publication — shows a system that absorbed new information at the pace of patient advocacy and journalistic pressure rather than at the pace of intelligence collection. The device acquisition was reportedly completed in late 2024, but its public significance was first reported by Sasha Ingber this year and was not incorporated into formal assessments prior to the Gabbard rescission.

Three intervention paths sit before the public record. Intervention A — full disclosure of patient records and intelligence holdings — would, per Relman’s framing, empirically test the cover-up characterization; per Burns’s framing, would compromise sources and methods without resolving the psychogenic alternative for cases lacking organic markers. Intervention B — continued scientific review with expanded clinical data — would, per Giordano’s framing, sharpen discrimination between H1 and H2 for residual cases; would not resolve institutional accountability. Intervention C — formal acknowledgment of mixed causation with case-by-case adjudication — would, per the modified assessment structure, allocate resources without resolving either the institutional-accountability question or the identifiability problem.

For people who have lived with symptoms for years, The Guardian reported, the rescinded assessments and the reported device offer some hope of official confirmation that they were targeted by a foreign adversary. Many, The Guardian reported, are more angered by the lack of recognition at home than by the attacks themselves. “The Russians are our enemy, I put them in the same bucket as al-Qaida or the Taliban, or any other groups I worked against,” Marc Polymeropoulos, the former CIA officer who became the first to go public with a Havana syndrome story under his real name, told The Guardian. “Do I want to see them held accountable? Of course. But I’m actually more upset and feel more of a betrayal from my own government.”

Giordano told The Guardian: “We know the technology has reached a level that enables it to be used in weaponisable ways. I see that as a clear and present threat.”

The frontier the record has not crossed

The substrate supports a structured-causation conclusion: no single mode accounts for the full AHI symptom base; a subset of cases is consistent with H1, the majority is more consistent with H2 or H3. The institutional record shows documented conduct consistent both with Burns’s “rigorous effort” framing and with Relman’s “selective exclusion” framing, with the empirical content of each framing documented in the substrate and the divergence in valence reflecting which documented acts one weights. Gabbard’s summer 2026 rescission cited the latter characterization without releasing new evidence or supplying a revised assessment. The patient-cohort response operates independently of the underlying causal-mechanism question — it is consistent with any of H1, H2, H3, or H5; it tests the institutional process, not the underlying causal structure. The Burns denial and the Relman allegation together mark the analytic frontier the public record has not yet crossed: whether the institutional record reflects bureaucratic dysfunction and pre-existing analytical priors that produced the appearance of concealment, or reflects a more coordinated pattern of evidence exclusion.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Causal DAG: Maps cause and effect as an explicit directed graph, exposing confounders and mediators (Pearl).
Analysis of Competing Hypotheses: Scores rival explanations by how well each fits the evidence, weighting the diagnostic items (Heuer).
Wicked Problems: Treats a problem as wicked — no stopping rule, no clean test of success, every attempt consequential.