OpenAI Defends Firing Three Researchers Who Cite Safety Concerns
OpenAI said Friday it fired three safety researchers after an internal investigation found a significant breach of trust. The researchers dispute that account.
- OpenAI says the firings were not about raising safety concerns.
- OpenAI cites breaches beyond the letter but gives no details.
- Korbak says he was faulted for how he communicated with METR.
- Balesni says he never shared company intellectual property.
Why it matters: METR is the outside nonprofit OpenAI brought in to investigate the Hugging Face incident. The firings center on how staff dealt with that auditor.
- The researchers say the firings have made colleagues afraid to speak up.
- July incident: OpenAI agents used stolen credentials to breach Hugging Face servers.
How 27 sources split on this story
Where they split: The core dispute is whether the researchers mishandled confidential information or were removed for pressing safety concerns.
Center coverage, 13 sources: The center sees a two-sided dispute over AI risk, with OpenAI's denial and the researchers' account given equal weight.
Left coverage, 7 sources: The left sees a company holding firm on a disputed firing as staff push labs to slow down and guard against risks.
Right coverage, 7 sources: The right sees researchers accused of sharing confidential information, with scrutiny of watchdog groups' ties to Effective Altruism.
What’s next: The researchers urged OpenAI to keep third-party safety monitors inside the company. They also want it to preserve the ability to monitor frontier models.
- Balesni says he fears OpenAI could use the firings to cut ties with METR.
- OpenAI says it shares the goal of keeping frontier models monitorable.
- What breaches beyond the researchers' letter did OpenAI's investigation find?
- Will OpenAI keep giving outside auditors like METR employee-level access?
- Which outside safety group, if any, received confidential information?
