Emil Michael, the Pentagon official overseeing U.S. military artificial intelligence policy, sold his holdings in Perplexity — the AI search company reportedly seeking a $30 billion valuation — for between $5 million and $25 million, according to financial disclosures examined by The Guardian.
The sale, executed in June, comes on top of earlier trading that has made Michael one of the wealthiest officials in the building. In January, The Guardian reported he reaped profits of up to $24 million selling his private investment in Elon Musk's xAI, a windfall representing gains of 400% to 4,800% on his original stake. Federal financial disclosures report dollar values in ranges rather than exact figures, so neither sale's precise proceeds are known. For more context on this story, see our ongoing latest AI developments.
Deep Ties to the Company He Traded
Michael's connection to Perplexity long predates the sale. As The Lever reported in March, he held a stake in the company. His 2025 disclosures show he also received a personal loan from Perplexity a year earlier — between $250,000 and $500,000 at a 4.57% interest rate — and had joined the company's advisory board, stepping down only upon entering federal service.
Some of his Perplexity stock was vested and some unvested, and his ethics agreement pledged he would not profit from the unvested shares. The disclosures do not specify which stock he sold in June, nor whether the transaction produced a profit.
Perplexity's government footprint adds context: the company announced a contract with the General Services Administration in November of last year, though the records reviewed do not show specific ties between the company and the Pentagon itself.
Ethics Experts Flag the Timeline
Richard Painter, who served as the White House's chief ethics lawyer under President George W. Bush, told The Guardian Michael should have divested before taking office. "He should have sold all interest in the company before he started working," Painter said, adding that under most administrations "we would have said get out of that stock before you start that job." Painter noted that holding stock in an AI company while shaping AI policy would not necessarily violate the law — but raises questions of appearance.
The Pentagon pushed back forcefully. A department spokesperson said Michael and other defense officials "are in full compliance with ethics laws and regulations," called contrary claims false, and pointed to the department's "rigorous, multi-layered ethics framework."
The Trading Extends Beyond AI
Michael's June Perplexity sale is not the only large transaction on his latest disclosure. The records show he sold holdings in Brex LLC — the Peter Thiel-backed financial software company acquired by Capital One in April — for at least $5 million, realizing gains of at least 473%. Michael had worked as a consultant at the firm, and as recently as his March 2025 disclosure he valued those holdings at a maximum of $750,000. His profit on the Brex sale could range from $4.25 million to $24 million, the figures show.
Unlike xAI, Brex does not appear to do reported work with the U.S. government.
The Man Setting Military AI Policy
Michael's portfolio of responsibilities makes the optics acute. As the official overseeing military AI policy, he has been described as the face of the Pentagon's confrontational posture toward Anthropic, one of the country's leading AI labs. Before government, Michael was chief business officer at Uber from 2014 to 2017, where his tenure ended after BuzzFeed News reported he had privately floated digging up dirt on critical journalists — a comment he said, made at what he believed was an off-the-record dinner, did not reflect his views.
The Guardian's reporting was picked up by Yahoo Finance on Tuesday under the headline "Top Pentagon AI Official Sold Millions in Perplexity Stock, Disclosures Show," broadening the story's reach as Washington's AI policy apparatus continues to expand.
A Timeline of Entanglement
Seen together, the disclosures sketch a sequence that ethics researchers will parse for months: an advisory board seat resigned at the doorstep of federal service, a personal loan from the company recorded in disclosures a year before the sale, unvested shares carried into office under an ethics pledge, and then a market-timing sale executed in June — all while the official held policy responsibility over the very sector in which the company competes. Each step has its own documented explanation; the cumulative picture is what has drawn the criticism.
Why It Matters
The episode lands amid an unresolved debate in Washington over how officials who steer AI policy should manage their ties to the industry they regulate — an issue that spans stock ownership, advisory roles, and the revolving door between labs and government. Federal disclosure forms capture the transactions after the fact, in ranges, months late; they are not designed to reassure the public in real time.
For the AI industry, the stakes are concrete: Pentagon procurement choices, positions on frontier-lab disputes, and the department's posture toward companies like Anthropic and its rivals all flow through officials like Michael. Whether his trading was lawful is not seriously contested. Whether it should have been permitted at all is the question ethics experts say will outlast the disclosures.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
