OpenAI fired three employees after incidents with confidential data
OpenAI fired three employees related to safety and alignment of artificial intelligence (alignment), after an internal investigation into the handling of confidential corporate information.
The Wall Street Journal was the first to report the dismissals. Later, OpenAI itself confirmed the information: the company stated that the investigation revealed violations of established procedures for accessing and handling sensitive information.
According to The Information, one of those fired worked directly on AI safety, the second on alignment, and the third was a research program manager in this area. OpenAI stated that it was not a single case, but a series of violations of the rules for handling research information.
A person familiar with the circumstances of the case told the media that one of the episodes involved the transfer of sensitive information to a third-party organization engaged in independent evaluation of artificial intelligence systems. OpenAI did not disclose the name of this organization.
According to WSJ, these are Jasmine Wang, Tomek Korbak and Mikita Balesni. OpenAI itself did not publicly name the dismissed employees. Recently, all three worked or were associated with issues of safety and risks related to the development of powerful AI models.
At the same time, the fact of OpenAI's interaction with external organizations that test the safety of models is not unusual. The company and other AI developers involve independent experts to assess risks. However, access of such teams to internal materials is regulated by special procedures and restrictions.
The dismissals occurred amid increased attention to the safety of OpenAI products. In recent months, the company has published results of investigations into incidents involving the behavior of autonomous AI agents and has strengthened internal procedures for monitoring potentially dangerous capabilities of models.
Based on materials: The Wall Street Journal, The Information
