Content
summary Summary

Further details of Altman's 4.5-day dismissal from OpenAI have been leaked.

Ad

Altman is said to have pitted members of the board against each other to force Helen Toner, with whom he had a dispute over a paper, off the board. This was reported by The New Yorker's Charles Duhigg, citing a source familiar with the events.

Decoding Intentions

Toner co-authored the paper "Decoding Intentions," which criticizes the AI hype generated by ChatGPT and the "race to the bottom" in AI safety that it may have triggered.

At the same time, the paper praises OpenAI's competitor Anthropic, which initially decided against releasing its chatbot Claude and only relented after ChatGPT was published, according to the paper.

Ad
Ad

"By delaying the release of Claude until another company put out a similarly capable product, Anthropic was showing its willingness to avoid exactly the kind of frantic corner-cutting that the release of ChatGPT appeared to spur," the paper says.

Unlike OpenAI, Anthropic chose not to commercialize its technology to avoid further fueling the AI hype, the paper states.

Altman reportedly manipulated the board

The paper's accusation that OpenAI had deliberately fueled the AI hype led to a confrontation between Altman and Toner. Altman then allegedly plotted against Toner and held one-on-one meetings with the board to replace Toner.

When the board members talked to each other, they realized that Altman had misrepresented some of the board's statements as supporting Toner's ouster. The source said that Altman tried to play the board members off against each other.

Altman himself allegedly described his approach as "ham-fisted", but not manipulative. Some board members lacked experience and were intimidated by their responsibilities. He viewed the discussions as a "very normal and healthy boardroom debate."

Recommendation

Ultimately, the board felt that Altman was no longer the trustworthy and honest CEO that a safety-conscious AI company like OpenAI needed. Trust in Altman was lost.

The board feared Altman's power, so the firing had to be quick

Altman is known internally as a tough negotiator who has repeatedly prevailed in confrontations in recent years, including with Elon Musk in 2018 when he tried to acquire OpenAI - and then backed out.

So when the board discussed Altman's ouster, Toner, D'Angelo, Sutskever, and Tasha McCauley agreed that it had to be done quickly and take Altman by surprise. If Altman got wind of his removal beforehand, he would do everything he could to prevent it.

Then things took their course.

Ad
Ad
Join our community
Join the DECODER community on Discord, Reddit or Twitter - we can't wait to meet you.

Although Altman returned to his old position after just 4.5 days under pressure from Microsoft and the staff, and Toner lost his role on the board, she considers the move a success: "Sam is very powerful, he’s persuasive, he’s good at getting his way, and now he’s on notice that people are watching."

Altman says he looks forward to an internal investigation into what went wrong during those few dramatic days that made tech history.

Support our independent, free-access reporting. Any contribution helps and secures our future. Support now:
Bank transfer
Summary
  • OpenAI's former and current CEO, Sam Altman, allegedly tried to remove board member Helen Toner after she co-authored a paper critical of AI hype and OpenAI.
  • Altman was accused of holding one-on-one meetings with board members and manipulating their comments to force Toner off the board.
  • Despite his brief ouster and later reinstatement as CEO, Toner is satisfied that Altman now knows he is under scrutiny, she says.
Sources
Online journalist Matthias is the co-founder and publisher of THE DECODER. He believes that artificial intelligence will fundamentally change the relationship between humans and computers.
Join our community
Join the DECODER community on Discord, Reddit or Twitter - we can't wait to meet you.