OpenAI fired three employees working on its AI safety teams for improperly disseminating sensitive internal company information. According to The Wall Street Journal, the employees shared the information with a third-party AI safety organization.
OpenAI said the employees had undermined internal trust by violating the organization’s policies on accessing and using sensitive information. While the content of the shared information, the name of the organization and the method of transfer were not disclosed, it was stated that those fired included two security managers and one research program manager.
Why it matters
This development shows that, for teams working in the field of AI safety, rules governing access to and use of information are at the center of corporate responsibility alongside technical expertise. The involvement of security and research program managers means the matter needs to be assessed not only as an individual disclosure, but also in terms of internal oversight and trust mechanisms designed to prevent the transfer of sensitive data outside the organization. However, since the information shared, the identity of the third-party organization and how the transfer took place were not disclosed, no definitive assessment can be made of the scope of the incident or its potential effects. These uncertainties leave open questions about how OpenAI’s access policies are implemented and what processes will be put in place to prevent similar situations.
Background
OpenAI is not a new name in the FikirPilot archive: we have published 75 news reports mentioning the name in the past 90 days; the most recent was dated October 3, 2026.