OpenAI Safety Firings Spark Chilling Effect Warning
Three safety researchers fired from OpenAI last week are pushing back hard. Jasmine Wang, Tomek Korbak, and Mikita Balesni published an open letter disputing the company's misconduct claims and warning that their abrupt dismissals are already scaring colleagues away from speaking up. OpenAI says the firings followed an investigation into mishandling sensitive information. The researchers say they were doing exactly the kind of external collaboration that safety work requires.
What OpenAI Says Happened
OpenAI fired the three researchers after an investigation found a pattern of misconduct in clear violation of policies on mishandling research information, according to a spokesperson. An internal memo attributed to a research leader praised their contributions to AI safety and denied retaliation, saying the decisions were not about raising safety concerns. OpenAI did not answer TechCrunch's questions about which specific policies were violated, the circumstances of the dismissal, or how the company protects employees who raise concerns and work with external evaluators. The company has not formally responded to the open letter.
The Researchers' Defense
The three deny mishandling sensitive information and deny leaking to The Information about less monitorable architectures in OpenAI's newest models. They also deny engaging with external parties outside their job mandates. Wang says she was fired for accessing an executive's email, but that OpenAI delegated that access to her for recruiting, she asked IT to remove it, and she reported opening a sensitive email by mistake within minutes. Korbak believed he was acting within company norms by communicating with outside safety evaluators during the unprecedented Hugging Face incident investigation. Balesni coordinated with board members and executives on monitorability work, the letter says, and removed sensitive details before sharing materials. The researchers say their actions were in good faith and within the norms as they stood at the time.
Why This Matters Beyond OpenAI
The letter argues that freedom to collaborate with outside experts, without fear, is itself an essential safety mechanism. If safety researchers at the world's most prominent AI lab are afraid to speak or work with external evaluators, the chilling effect ripples across the whole ecosystem. The researchers call on OpenAI to honor public commitments to embed third-party safety auditors, preserve monitorability of frontier models, and support open dialogue between safety researchers and the broader safety community. OpenAI says it agrees with these recommendations. But the dispute raises a harder question for every AI company: when safety work requires external collaboration, where exactly is the line between diligence and dismissal?
Key Takeaways
- Three fired OpenAI safety researchers deny misconduct claims and warn of a chilling effect on AI safety culture.
- OpenAI says an investigation found a pattern of policy violations in mishandling research information.
- The researchers call on OpenAI to embed third-party auditors and preserve frontier model monitorability.
- The dispute highlights tension between internal confidentiality and external safety collaboration across the AI industry.
Source: TechCrunch • 🇺🇸 San Francisco
Keep Reading
