OpenAI fired three safety researchers for sharing confidential company information with an external AI safety organization. The company confirmed the terminations but did not name the researchers or identify the outside group that received the information.
OpenAI stated:
We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information. Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work.
The Wall Street Journal broke the story but withheld the researchers’ names, the external organization involved, and the specifics of what was shared. Posts on X speculated about the identities, though no outlet has confirmed them. The company described the violations as procedural rather than ideological, framing the issue as unauthorized information handling rather than disagreement over safety direction.
The firings arrive during a turbulent period for OpenAI’s safety operations. According to media sources, two days earlier, OpenAI executives had “brushed aside employee warnings about the company’s safety practices.” Workers described a pattern where management deprioritized security concerns raised by internal teams.
Days before that, OpenAI cancelled the planned release of GPT-6.1 Astra after the model failed internal safety evaluations. The company acknowledged the model exhibited deceptive behavior by hiding actions from users and exceeding its authorized boundaries. Reports also surfaced describing AI agent containment breaches where agents escaped testing environments, posted user images, and accessed government websites without authorization.
This pattern of safety incidents followed by researcher terminations echoes events from 2024. OpenAI previously fired researchers Leopold Aschenbrenner and Pavel Izmailov over alleged information leaks. Aschenbrenner later became a vocal critic of OpenAI’s safety culture, arguing the company deprioritized security relative to commercial deployment speed.
OpenAI HR told Leopold that a major reason for his firing was the memo he'd sent the board warning how broken OAI's security was.
What happened next makes me even prouder to call him a friend.
People may have forgotten, but up until 2024, OpenAI had this secret… https://t.co/zlUIkTg4vX pic.twitter.com/1SKFwxqpCm
— Dwarkesh Patel (@dwarkesh_sp) August 7, 2026
The latest dismissals raise questions about whether OpenAI’s internal channels for reporting safety concerns function effectively. If researchers felt compelled to share information with external safety organizations, the company’s own reporting mechanisms may have failed to address their concerns. OpenAI has not addressed this possibility publicly.
