
AI labs are hiring spy hunters: Anthropic 33, OpenAI 22, Nvidia 16 counterintelligence staff in 2026
By Himanshu Sakre·
Revelio Labs data reported by Fortune and Reuters shows the leading AI companies adding counterintelligence staff at pace, after an alleged OpenAI office bugging and Nvidia's leadership scandal. The counts have a definition problem.
The people AI companies are hiring this year include former spies. Fortune reported on November 14, 2026 that Anthropic has made 33 counterintelligence hires so far in 2026, up from 14 in 2025 and nine in 2024, citing Revelio Labs data compiled by Reuters. OpenAI is at 22 for 2026, up from nine in 2025, and Nvidia at 16.
The trend has a news hook that is hard to miss. Nvidia's chief executive resigned this month amid a scandal that involves an alleged Chinese spy, and an OpenAI engineer was charged in October over a hidden recording device. The numbers deserve a careful read, though.
The cases behind the hiring
According to the Justice Department's complaint as reported by Al Jazeera on October 13, 2026, OpenAI engineer Abdulrahman Alfozan, 35, is charged with using or attempting to use a device to intercept oral communications. The maximum penalty is five years in prison and a $250,000 fine. Prosecutors allege a device was hidden behind a television in an executive conference room, installed around September 28 and discovered on September 30. These are allegations, and Reuters reports he pleaded not guilty on November 6.
OpenAI's statement, as quoted by Al Jazeera, said: "we have no indications of this individual inappropriately accessing or exfiltrating model weights." The company said it referred the matter to the FBI. FBI Director Kash Patel said the bureau treats protecting American innovation as "one of our highest counterintelligence priorities."
The Nvidia episode is murkier. Reuters reports that Jensen Huang admitted one of two reported extramarital affairs, one involving an alleged Chinese spy, and that Nvidia said it did not believe national security or company secrets were compromised. China rejected the espionage allegations. Our report on the House briefing covers what lawmakers asked for.
What the counts do and do not say
Fortune presents the Revelio figures as hires. Reuters' piece distinguishes between job ads and named appointments. It lists open roles: OpenAI is hiring a senior global security investigator, with counterintelligence a key skill; Anthropic seeks a threat-intelligence security engineer; ASML advertises an intelligence officer in Taiwan requiring at least 10 years of counterintelligence experience. It also lists named hires: former NSA director Paul Nakasone joined OpenAI's board and safety committee in June 2024, and Anthropic added former CIA deputy director David Cohen and former NSA cybersecurity director Dave Luber to its national security advisory council in August 2025.
“Demand outstrips the supply of people with relevant experience.”
We could not open Revelio's methodology. So a count of 33 may reflect people hired into counterintelligence-flavored roles, or a classification of job titles. Either way, the direction is consistent across both outlets: more of these roles, at higher pay.

Pay and the talent squeeze
Reuters reports Revelio says counterintelligence roles at frontier AI firms pay about $200,000 to $500,000. Recruiter Kathy Lavinder told Reuters top corporate security chiefs can earn $1 million to $4 million a year, depending on stock awards, and that "Security teams don't have a veto." Christopher Falkenberg, president of Insite Risk Management and a former Secret Service agent, said: "Demand outstrips the supply of people with relevant experience."
Fortune adds that the Washington Post reported in October a Pentagon plan for covert officers posing as executives, and that Politico reported on October 22 that the Defense Department is building an agency to covertly sabotage foreign technology programs. Fortune says the Pentagon has denied the reports, and we have not verified either claim.
Why this belongs in the race
Frontier labs hold model weights, training recipes and compute plans that rival states want. Anthropic's own advisory council appointments and OpenAI's board choices show how closely labs now track national security expertise. The same few companies also sit at the center of the compute and financing story, which raises the value of what they hold.
What governments are doing
Fortune's account of the government side is wider than hiring. It says CIA Deputy Director Michael Ellis said in September 2026 that the United States is expanding espionage against China, targeting companies in AI, biotechnology and chips. It reports that the Pentagon formed an Economic Defense Unit in April 2026, citing DefenseScoop, and that Politico reported a Technology Support Office, set up in April 2026 by reorganizing two classified programs that had run technology sabotage at a smaller scale since 2023. The Pentagon has denied the sabotage reports, per Fortune.
Former FBI counterintelligence operative Eric O'Neill told Fortune that executives need to learn to recognize manipulation, and listed warning signs: relationships that move unusually fast, persistent questions about sensitive business, requests to meet influential colleagues, and encouragement to keep secrets. Approaches, he said, can begin at conferences, business introductions, over drinks or on dating apps.
Reuters adds an industry angle. ASML, Europe's most valuable listed company, set up an Intelligence Fusion Centre, and its former security chief estimated that 90 percent of companies ignore the risk of executives being targeted. The chip supply chain, not only the labs, is now a counterintelligence perimeter.
Our take
Hiring spy hunters is a sensible response to a documented case and a documented scandal. It is not evidence that weights have been stolen. In the one prosecuted case, OpenAI says it has no indication of weights being taken.
We would watch for two things: whether Revelio or Reuters publishes the definition behind these counts, and whether any lab discloses what its counterintelligence teams have found. Until then, treat the headline numbers as a trend line, not a ledger.
Frequently asked questions
How many counterintelligence staff have AI companies hired in 2026?
Fortune reports, citing Revelio Labs data compiled by Reuters, 33 at Anthropic, 22 at OpenAI and 16 at Nvidia so far in 2026. Anthropic was at 14 in 2025 and nine in 2024.
Why are AI labs hiring counterintelligence experts?
Reuters and Fortune tie the trend to concerns about theft of trade secrets and espionage, with Nvidia's leadership scandal and the charged OpenAI bugging case as recent examples.
What is the OpenAI bugging case?
The Justice Department charged OpenAI engineer Abdulrahman Alfozan with using a device to intercept oral communications after a recording device was found in an executive conference room on September 30. These are allegations, and he pleaded not guilty on November 6.
Did the OpenAI case involve stolen model weights?
OpenAI said it had no indications of the individual inappropriately accessing or exfiltrating model weights, as quoted by Al Jazeera.
How much do counterintelligence roles at AI firms pay?
Reuters reports Revelio Labs says about $200,000 to $500,000 for most such roles at frontier AI firms. A recruiter said top security chiefs can earn $1 million to $4 million a year.
Are the hiring counts reliable?
They are indicative. Fortune calls them hires, Reuters separates job ads from named appointments, and we could not open Revelio's classification method.
Sources
What each one is, and whose it is.
- 1
The hottest job in AI right now may be counterintelligence, Fortune (November 14, 2026)
Press reportIndependent of the vendor - 2
Help wanted: AI firms court ex-spies as Nvidia drama spotlights security risks, Reuters via WMBD (November 12, 2026)
Press reportIndependent of the vendor - 3
OpenAI engineer charged after listening device found in meeting room, Al Jazeera (October 13, 2026)
Press reportIndependent of the vendor