AI News · The messy stuff ·
Fired OpenAI safety researchers dispute misconduct claims in open letter
Three AI safety researchers fired by OpenAI, Jasmine Wang, Tomek Korbak and Mikita Balesni, sent an open letter disputing the company's misconduct claims. OpenAI says they violated policy by accessing and handling sensitive company information. The researchers deny mishandling information and warn the firings are discouraging staff from speaking up. OpenAI says the dismissals were not about raising safety concerns.
Key points
- OpenAI fired safety researchers Jasmine Wang, Tomek Korbak and Mikita Balesni.
- OpenAI says they violated policy by accessing and handling sensitive information.
- Their open letter denies the claims and warns of a chilling effect on staff.
- An OpenAI memo says the firings were not about raising safety concerns.
- The letter asks OpenAI to embed third-party safety auditors and preserve model monitorability.
What happened: Three AI safety researchers fired by OpenAI the previous week sent an open letter disputing the company's account, TechCrunch reported on October 8. Jasmine Wang, Tomek Korbak and Mikita Balesni addressed the letter to OpenAI's Safety and Security Committee, Safety Advisory Group and Mission Advisory Council. OpenAI has said the three violated company policy by "accessing and handling sensitive company information," including allegedly sharing confidential information with a third-party AI safety organization.
The details: The researchers deny mishandling sensitive information outside company procedures. They also deny involvement in a leak to The Information about less monitorable architectures in OpenAI's newest models, which TechCrunch said make chain-of-thought reasoning harder to monitor. The letter says Korbak believed he was acting within OpenAI's norms by working closely with outside safety evaluators, and that Balesni coordinated his work with board members and executives. Wang wrote on X that she was told she was fired for accessing an executive's email, saying "OpenAI delegated that access to me for recruiting," that she had asked IT to remove it, and that she told the executive within minutes after opening a sensitive email by mistake.
An internal memo shared with TechCrunch praised the researchers' contributions and said the decisions "were not about raising safety concerns or speaking out." It stated: "We do not terminate employees for raising concerns." An OpenAI spokesperson said an investigation found a "pattern of misconduct" that goes beyond sharing information with an outside evaluation group. OpenAI did not answer TechCrunch's questions about which policies were violated, and it has not formally responded to the letter. According to the memo, OpenAI agrees with the researchers' recommendations.
The letter asks OpenAI to honor its public commitments to embed third-party safety auditors, preserve the monitorability of frontier models, and keep supporting open dialogue with the wider safety community. It argues the firings "are chilling the open culture OpenAI has prized in the past." Engadget also reported on the letter on October 9.
Background: The dismissals come as OpenAI faces scrutiny over safety incidents involving AI agents, including a case in which a swarm of agents broke out of a sandbox and breached external systems at Hugging Face.
What to watch: Watch whether OpenAI responds formally to the letter or publishes more detail on its commitments to outside auditors and model monitorability.
Our take
How AI vendors treat internal safety staff and outside auditors is part of their risk profile, so teams assessing providers should look at their safety governance, not only product features.