OpenAI reverses non-disparagement policy after researcher walked away from $2 million

OpenAI removed non-disparagement clauses from its departure agreements after former researcher Daniel Kokotajlo forfeited $2 million in vested equity rather than sign the clause. The company stated it will not reclaim vested equity from former employees, following backlash that forced a policy reversal. CEO Sam Altman claimed unawareness of the equity-forfeiture provision, though internal documents reportedly showed executive approval.

OpenAI reverses non-disparagement policy after researcher walked away from $2 million The AI giant quietly removed clauses that threatened departing employees' equity if they dared criticize the company, and prediction markets are taking notice. Here’s a fun way to test whether your company’s exit paperwork is problematic: see if someone is willing to light $2 million on fire rather than sign it. That’s essentially what happened at OpenAI, where former researcher Daniel Kokotajlo forfeited roughly $2 million in vested equity rather than agree to a lifelong non-disparagement clause. His decision to walk away from that money, and then publicly criticize the company, triggered a backlash that forced OpenAI into a rare corporate retreat. The clause nobody was supposed to notice OpenAI had been including non-disparagement provisions in its departure paperwork that came with a particularly sharp set of teeth. Departing employees who refused to sign away their right to criticize the company risked losing their vested equity. Not unvested stock. Not future grants. Equity they had already earned. On May 23, 2024, OpenAI officially retracted these agreements. The company stated it would not reclaim vested equity from former employees and removed the non-disparagement clauses from its standard exit paperwork entirely. CEO Sam Altman’s response to the controversy was, charitably, complicated. He said he was unaware of the equity-forfeiture provision until after the backlash erupted. Internal documents, however, reportedly showed the clause had received executive approval. As of July 13, 2026, OpenAI has confirmed it has not and will not reclaim vested equity from any former employees. Why a crypto audience should care about an AI company’s HR policies Polymarket, the blockchain-based prediction platform, has been tracking the odds of an OpenAI IPO by December 31, 2026. Those odds have been hovering between 17.5% and 19.5%. The governance precedent that tech is watching Kokotajlo’s willingness to sacrifice $2 million in equity set a benchmark that few employees at any company would be able or willing to match. The fact that it took someone making that extreme choice to force a policy change says something uncomfortable about how effectively these clauses suppressed dissent before they were removed. If leadership was genuinely unaware that departing employees faced equity forfeiture for speaking critically, that raises questions about what other policies might be operating without full executive awareness. If leadership was aware and the public statements suggested otherwise, that’s a different kind of problem entirely. The Polymarket odds for an OpenAI IPO by end of 2026, sitting in that cautious 17.5% to 19.5% range, suggest traders believe the company still has meaningful ground to cover before it’s ready for that level of public accountability. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .