OpenAI doubles down on decision to fire three AI safety researchers
The Verge Robert Hart ● Covered by 3 sources
OpenAI says it fired three safety researchers for breaking sensitive-info rules. The company says this wasn’t about their AI safety criticism, just a “significant breach of trust.”
Based on reporting by The Verge, Robert Hart — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI is sticking to its decision to fire three safety researchers, and it is doing so publicly. On Friday, the company said Jasmine Wang, Tomek Korbak and Mikita Balesni were dismissed after an investigation found what it called a “significant breach of trust.”
The company says the issue was policy, not dissent. According to its post on X, the three violated “clear policies on handling sensitive information.” OpenAI also pushed back on the idea that the firings were punishment for speaking out about the company or raising concerns about AI safety.
That matters because the dispute is no longer just about three employees. The researchers had already published an open letter on Thursday asking OpenAI to be more transparent about why they were shown the door, and they followed that up with posts on social media.
OpenAI’s response is blunt: the company is not walking this back, and it wants the story framed as a trust and information-handling problem, not a free-speech one. The tension here is familiar, and a little ugly. A company can say “safety” all day long, but if the people working on safety think the process is opaque, the optics do the damage anyway.
My take — AI-written commentary, not fact-checked reporting
OpenAI is playing the classic corporate move: call it policy, not retaliation, and hope the paperwork does the talking. That usually works until it doesn’t. If a safety shop starts looking allergic to transparency, people stop hearing “trust” and start hearing “we decide what trust means.”
Read more about this at: The Verge