A newly launched court deposition reveals the chaotic days surrounding Sam Altman’s 2023 ouster from OpenAI, and reveals simply how shut the corporate got here to merging with its prime rival, Anthropic.
The sworn testimony is from OpenAI co-founder and former chief scientist Ilya Sutskever, who spoke beneath oath as a part of Elon Musk’s lawsuit towards OpenAI’s Sam Altman.
The deposition:
- Particulars a failed plot to merge with Anthropic
- Confirms the existence of a 52-page memo accusing Altman of manipulation
- Reveals a second, related memo was written about co-founder Greg Brockman
To dig into this on-the-record testimony, I talked with SmarterX and Advertising AI Institute founder and CEO Paul Roetzer on Episode 179 of The Artificial Intelligence Show.
The Anthropic Merger That Nearly Was
Sutskever’s deposition confirms that within the instant aftermath of Altman’s firing, talks started to merge OpenAI with Anthropic, a transfer that may have probably put in Anthropic’s CEO, Dario Amodei, on the helm.
Anthropic initially expressed “pleasure” concerning the potential deal, in accordance with Sutskever.
The talks shortly collapsed, nevertheless, attributable to what Sutskever described as “sensible obstacles,” probably associated to the advanced investor agreements each corporations held. Sutskever additionally described “deep distrust” inside OpenAI’s management.
He additionally testified that different board members had been much more supportive and recognized former board member Helen Toner as suggesting the merger.
Not True, Says Toner
Toner, nevertheless, shortly took to social media after the deposition was launched to dispute this. She tweeted:
“For the report, for these dissecting Ilya’s deposition: This half is fake. I wasn’t the one who made the board <> Anthropic name occur, and I disagree along with his recollection that board members aside from him had been supportive of a merge.”
Roetzer notes that her public correction provides one other layer of complexity to the competing narratives.
The 52-Web page Altman Memo
Past the merger drama, the deposition offered gorgeous, on-the-record affirmation of the 52-page memo Sutskever wrote detailing his case towards Altman.
He despatched the memo solely to the board’s impartial administrators, testifying that he deliberately saved Altman in the dead of night. Why?
“As a result of I felt that, had he grow to be conscious of those discussions, he would simply discover a option to make them disappear,” Sutskever says.
The memo was damning. Its very first web page, learn into the court docket report, acknowledged: “Sam displays a constant sample of mendacity, undermining his execs, and pitting his execs towards each other.”
When requested what motion he believed was acceptable primarily based on that memo, Sutskever’s reply was: “Termination.”
A Second Memo and Extra Drama
The testimony additionally revealed that Sutskever’s memo was constructed, partly, on proof provided by others, together with then-OpenAI Chief Know-how Officer Mira Murati. Sutskever testified that lots of the screenshots documenting Altman’s conduct got here from Murati.
However maybe the most important shock was the revelation that Altman wasn’t the one goal. Sutskever confirmed he additionally drafted an analogous memo essential of OpenAI co-founder (and present President) Greg Brockman.
Your entire saga, from secret memos and government infighting to a close to rival merger, paints an image of inside battle much more intense than beforehand recognized.
“If anyone watched The Social Community and the Fb loopy story, that is it on steroids,” Roetzer says. “The OpenAI film goes to be insane.”
