OpenAI and Sam Altman: How a Board Crisis Turned Into a Power Struggle

A column on OpenAI's internal crisis pulls several episodes into a single picture: the failed firing of Sam Altman, questions about his investment ties, and a new dispute over defense contracts. The author's central thesis is simple: after the crisis of 17 November 2023, the head of OpenAI returned not weakened but practically untouchable.

The November coup

On 17 November 2023, OpenAI's board removed Altman, saying he had not been "consistently candid" in his communications with the board. The wording immediately raised the key question: if this was not about a product mistake or about strategy, what exactly was the breach of trust? Within days the situation had turned into a power struggle. Employees demanded the CEO's return, Microsoft offered a landing place for his team, and influential Silicon Valley allies launched a campaign to reinstate him.

"Not consistently candid" — that is how the board explained the firing on 17

November 2023.

By 21 November Altman was back at the company, and the old board had effectively lost control. The author sees this as more than a personnel reversal: it was the moment OpenAI stopped being an organization with working internal checks. Even if the board's complaints were serious, it had neither the political resources nor the coalition to see its decision through. The crisis ended up strengthening the very person it had tried to remove.

Conflicts and structure

The second set of complaints concerns the overlap in Altman's roles as head of OpenAI, investor, and holder of many personal stakes in startups. In 2024, attention fell on the OpenAI Startup Fund: it emerged that the fund, associated with OpenAI, had been set up through a separate structure in which Altman held legal control. Later, on 29 March 2024, the structure was changed and formal control passed to Ian Hathaway, but the episode did not settle the question of how far the company head's influence extends.

• Former board member Helen Toner said the board did not know about the launch of ChatGPT in advance and learned of it only after release;
• Altman's individual investments in hundreds of companies fed concerns about conflicts of interest;
• critics also linked this to his influence over which startups get access to the OpenAI ecosystem;
• the later change to the fund's structure showed the issue was far from cosmetic.

There were also accounts of former board members coordinating through disappearing messages and trying to leave no digital trail. It sounds almost like a corporate thriller, but the author's point is different: if people at the board level feared speaking openly with their own CEO, OpenAI's governance was already deeply broken.

The contract and the reaction

A new wave of criticism was triggered by an episode in late February 2026 involving the Pentagon. Anthropic publicly refused to loosen restrictions on using its model for mass surveillance of citizens and fully autonomous weapons systems. Almost immediately, OpenAI signed its own agreement with the military, saying its "red lines" remained in place, although critics saw in the contract's wording the same risks Anthropic had tried to avoid.

Against this backdrop, the dispute over Altman moved beyond a corporate conflict and became a question of political accountability. Opponents see in his style not the ordinary toughness of a startup leader but a model in which speed, scale, and influence rank above transparency. Supporters reply that this approach is exactly what let OpenAI take the lead in the market. But public reaction showed that, for part of the audience, success no longer outweighs questions about who sets the rules and where moral limits end.

What this means

The OpenAI story looks less and less like a debate about one CEO's character and more and more like a test for the entire AI industry. If the company that sets the pace of the market can survive a crisis of trust without losing power at the top, then the main fight in the coming years will be not only over models but over who is capable of controlling their creators at all.