OpenAI Fires Safety Researchers Who Warned of AI Risk
If AI's own watchdogs are silenced, who keeps you safe?
OpenAI, the company behind ChatGPT, has fired three of its own safety researchers: Tomek Korbak, Mikita Balesni, and Jasmine Wang. OpenAI says they mishandled sensitive company information. But the researchers claim they were actually fired for advocating for AI safety—essentially for doing their jobs too well. They've published an open letter asking OpenAI to support open discussion and safety audits. OpenAI insists it does, but also says there was an additional breach of trust it won't explain.
Why should you care? These aren't random employees. They were central to efforts to keep advanced AI from going rogue. For example, they worked on ways to monitor AI's internal 'thoughts' to catch dangerous behavior before it happens. Jasmine Wang recently warned that we're speeding toward superintelligent AI that could be uncontrollable. Mikita Balesni estimated a greater than 10% chance AI could kill all humans. Tomek Korbak was a leading voice on alignment—making sure AI does what we actually want. Their firing removes key watchdogs from the company developing some of the world's most powerful AI.
The timing is troubling. OpenAI is under pressure to ship products fast, and safety concerns often take a backseat. If the company silences its own safety advocates, who will hold it accountable? The open letter and the researchers' statements suggest a culture where speaking up is risky. That's bad for all of us, because AI systems are already making decisions about hiring, loans, and even medical diagnoses. We need people inside these companies who can raise alarms without fear of retaliation.
This isn't just an internal drama. It's a warning sign. AI companies are largely self-regulated, and if they can't tolerate internal criticism, we can't trust them to keep us safe. The researchers' departures also chill free speech: other employees may think twice before voicing concerns. As AI becomes more integrated into our lives, we need transparency, not secrecy. The stakes are too high to ignore.
- Three OpenAI safety researchers were fired, allegedly for mishandling information, but they say it was retaliation for speaking up about AI risks.
- These researchers were key to efforts to monitor AI behavior and prevent disasters; their loss weakens internal safety checks.
- OpenAI won't fully explain the firings, raising concerns about secrecy and self-regulation in the AI industry.
Why It Matters
If AI companies silence safety experts, the technology you use daily could become riskier and less trustworthy.