(Credit: Jerod Harris / Stringer / Getty Images Entertainment via Getty Images)
Six months after OpenAI's board fired Sam Altman as CEO, former member Helen Toner is offering her side of the story in an interview on The TED AI Show podcast.
The board began to lose trust in Altman after "years" of him withholding information, misrepresenting OpenAI's activities, and in some cases "outright" lying to the board.
"We learned about ChatGPT on Twitter," says Toner.
Altman "made it hard" for the board to perform its role of "making sure that the company's public good mission was coming first over profits [and] investor interests." For example, he failed to disclose that he owned the OpenAI Startup Fund, "even though he constantly was claiming to be an independent board member with no financial interest in the company."
Safety was also a concern. Altman gave the board inaccurate information on "multiple occasions" about what guardrails the company had in place, making it "impossible to know how those processes were working or what needed to change," she says. (Two high-profile OpenAI execs recently quit over similar concerns; one of them landed at Anthropic.)
In December 2021, Toner published her own paper on AI safety, which included criticisms of OpenAI. Altman was angered by the paper and "started lying to other board members in order to try and push me off the board,” she says.
This went on for years, leading the four board members to conclude they "just couldn't believe" what Altman said. "That's a completely unworkable place to be as a board that is supposed to be providing independent oversight," Toner says.
Multiple OpenAI executives also expressed concerns about Altman, describing a toxic work culture and "psychological abuse." Some also questioned Altman's character and whether he could be trusted to lead efforts to crack artificial general intelligence (AGI). Screenshots and documents showed Altman lied and misconstrued important information, sealing the deal for the board in their decision to remove him, according to Toner.
In the interviews, Toner—who currently serves as Director of Strategy and Foundational Research Grants at the Center for Security and Emerging Technology (CSET)—highlighted Altman's previous work at a startup, where the management team asked the board to fire him for "deceptive and chaotic behavior." He then went to Y Combinator, where he was also fired.
Still, Altman's firing last year was met with intense pushback from OpenAI staff and major backers like Microsoft, and he was quickly reinstated.
When asked about the groundswell of employee support for Altman, Toner argued much of that was motivated by fear. "They've experienced him retaliate against people, retaliate against them for being critical. They were afraid," she says.
OpenAI employees also understood that his firing threatened the company as a whole, jeopardizing their financial futures.
After he returned to OpenAI in November—and Toner and Tasha McCauley were ousted from the board—Altman tweeted that "it is clear that there were real misunderstandings between me and members of the board."
More recently, when OpenAI named new board members, Altman again took to X/Twitter to say that "when I believed a former board member was harming OpenAI through some of their actions, I should have handled that situation with more grace and care. I apologize for this, and I wish I had done it differently." He didn't specify which board member.
McCauley expressed similar concerns with Altman in an op-ed she penned with Toner for The Economist this week.