A 'sociopath' at the helm of ChatGPT? A wide-ranging investigation paints a not-so-flattering portrait of OpenAI's CEO.

The New Yorker has just published the most detailed investigation ever devoted to Sam Altman. Signed by Ronan Farrow and Andrew Marantz, it is based on over 100 interviews. The journalists had access to the 'Ilya Memos' and 200 pages of private notes from Dario Amodei, the CEO of Anthropic, who was part of the founding team of the lab. The resulting portrait is devastating. It happens at the worst time: OpenAI is preparing an IPO that its own chief financial officer deems premature.

From Y Combinator to the 2023 ouster: a recurring pattern

The accusations are not new. Aaron Swartz, co-founder of Reddit and classmate of Altman at Y Combinator (the most popular startup incubator in Silicon Valley) in 2005, would have warned his circle. His words, cited by The New Yorker: Altman « can never be trusted » and would be a « sociopath ». Swartz died in 2013.

Altman had run Y Combinator for five years before being ousted. The accelerator claims he simply had to choose between his roles and OpenAI. Several sources from The New Yorker dispute this version.

At OpenAI, the pattern becomes clearer. Ilya Sutskever, then chief scientist, compiled internal memos totaling about 70 pages. The board produced a document detailing what it calls a « constant pattern » of Altman's lies. On November 17, 2023, the board dismissed him. The statement was brief: Altman had not been « consistently sincere in his communications ».

Five days later, Altman was back. He had mobilized his internal and external supporters. Most of the board members who had ousted him were replaced by allies: economist Larry Summers, former Facebook chief technologist Bret Taylor. OpenAI employees have rebranded the episode as "the Blip".

IPO, advertising, military: the choices that fuel distrust

Since his return, Altman has accelerated the transformation of OpenAI. The original non-profit structure has given way to a for-profit model. Internal security teams have been dismantled or reduced. The founding charter, according to sources cited by The New Yorker, no longer guides decisions.

The list of controversial turns is growing. Sora, OpenAI's video generator, has been abandoned. Advertising has made its way into the products. Contracts with the U.S. military sector have been signed. The company's valuation reached $852 billion during its last $122 billion fundraising round.

The New Yorker's investigation also documents tensions with Microsoft, a historic partner. Editors' executives describe Altman as someone who « distorts, renegotiates, and backs out of agreements ». A senior executive goes further. He estimates that Altman could one day be compared to « a Madoff or Bankman-Fried type scammer ». In parallel, an IPO is on the horizon, and Altman is pushing for the fourth quarter of 2026. The CFO Sarah Friar confided to colleagues that she didn't think the company was ready this year. She doubts OpenAI's ability to justify its massive AI server expenses. Revenue growth is slowing, and financial commitments are exploding. Altman has gradually distanced her from certain financial discussions.

Friar does not report to her directly since she has been attached to Fidji Simo, who is in charge of the applied branch.

OpenAI contested both investigations. On the New Yorker: « The article revisits already known events through anonymous testimonies and selected anecdotes, coming from people with clear agendas. » Regarding tensions with Friar, a joint statement with Altman states that both « were directly involved in every significant decision ».

The question raised by the investigation is not the one we think

The debate is not about Sam Altman's character. It is about the governance of a company developing technology that could reshape entire sectors of the economy and society. OpenAI employs thousands of people, manages billions of daily queries, and burns hundreds of billions of dollars in capital.

The independent investigation into the reasons for Altman's ouster in 2023 never produced a written report. The board that had judged him untrustworthy was completely replaced. The safeguards established at the creation of OpenAI were dismantled one by one. Indeed, this is the origin of the lawsuit between OpenAI and Elon Musk.

The man who wants to take OpenAI public is the same one that its own board tried to exclude less than three years ago. Investors will decide whether they trust him more than his former collaborators.

👉🏻 Follow real-time tech news: add 01net to your Google sources, and subscribe to our WhatsApp channel.

Source : The New Yorker