Sam Altman’s ouster and comeback at OpenAI plays out like corporate theater with boardroom coups, secret handshakes and vanishing Slack threads. When the board fired him, Altman set up a “government-in-exile” at his $27 million San Francisco home, pacing twelve-hour conference calls over nightly Negronis. Then board member Ilya Sutskever quietly compiled seventy pages of disappearing Slack messages to justify removing Altman and Greg Brockman—only to have much of that evidence hidden from public view. A follow-up probe led by Lawrence Summers and Anna Taylor never produced a written report, and some insiders still want a proper investigation. The New Yorker’s 18,000-word profile by Ronan Farrow and Andrew Marantz walks through these events, questioning Altman’s candor at every turn. It highlights his terse reaction to allegations—“I can’t change my personality,” he allegedly said when accused of deception—and his denial of key details, from the informal “shadow board” pact with Brockman and Sutskever to the surprise merger veto written into OpenAI’s charter for Microsoft. Paul Graham and Elon Musk both dispute versions of events, but the reporting leans toward Altman’s account on the Musk fallout and emphasizes the gulf between his public statements and board members’ private notes. Dario Amodei’s departure to found Anthropic also gets fresh context. He’d fought to keep a “merge-and-assist” safety clause in OpenAI’s charter; that protection vanished when Microsoft slipped in veto power over any merger. Amodei only discovered the change at the eleventh hour, forcing Altman to grudgingly admit it existed. Tensions escalated until Amodei and his sister Daniela walked away in 2020. Their new company now picks up the zero-day exploit work that OpenAI sidestepped—witness Anthropic’s cybersecurity partnership around its Claude Mythos model.