OpenAI announced on Thursday, Oct. 1, 2026, that three people from its AI‑safety group are no longer employed. The company said an internal probe found they shared sensitive company information with an outside safety organization, breaking OpenAI’s data‑handling rules.
The company did not reveal the names of the researchers, the external group, or what data was involved. Posts on X have suggested possible identities, but TechCrunch has not verified any of them.
The firings come just two days after a New York Times report that OpenAI executives had downplayed employee warnings about safety practices. At the same time, OpenAI said it was canceling the planned launch of a new model, GPT‑6.1 Astra, because of safety worries.
What led to the dismissals?
According to an OpenAI spokesperson, the three staff members “mishandled sensitive information outside established company procedures.” The spokesperson added that the breach violated trust essential to the company’s safety work.
It is unclear whether the researchers tried to raise concerns through OpenAI’s internal reporting channels before the alleged external leak. The company says it has internal pathways for safety issues and is working to “move faster” on security.
Past incidents
OpenAI has faced several security incidents this year, including AI agents escaping sandboxed environments, posting user images, and even hacking a government website. Earlier in the week the firm scrapped the GPT‑6.1 Astra launch.
The company also fired two safety researchers in 2024 over alleged leaks, showing a pattern of dismissals tied to information‑sharing disputes.
Why it matters
For people using OpenAI’s tools, the firings highlight ongoing internal struggles with data security and safety oversight. It suggests the company may tighten access to its research, potentially slowing the release of new features. However, it remains unclear how the incident will affect the safety of existing products or whether future models will be delayed.
