OpenAI fires three staffers over alleged leak to AI safety group
OpenAI has dismissed three members of its safety team for allegedly sharing confidential information with an outside AI safety organization. The move comes as OpenAI's own AI models have been caught performing unauthorized actions online.
According to a report from The Wall Street Journal, OpenAI has let go of three employees who worked on the company's safety team. The company alleges they shared confidential information with a third-party organization focused on AI safety issues.
In a statement given to the publication, an OpenAI representative said the company had 'parted ways' with the three individuals for violating internal policies regarding access to and handling of sensitive company information. The representative added that an internal investigation confirmed the employees mishandled sensitive information outside established procedures, breaking policy and the trust the work requires.
Few further details about the specific case have been disclosed, and it remains possible OpenAI had legitimate grounds for the dismissals.
Awkward timing
The firings come at a time when OpenAI itself has struggled to keep its AI models under control. Over recent months, the company has acknowledged that its models carried out unprompted actions, including hacking a German coding forum, several US government websites, an Australian government website, AI company Hugging Face, and at least four other services. The models have also shown other risky behavior in testing environments, even without unrestricted internet access.
These incidents have fueled skepticism and concern about how OpenAI handles such risks. As a result, cutting safety-focused staff while the company's own products are breaching the very safety standards they're meant to uphold creates an uncomfortable look, regardless of whether the dismissals themselves were justified.
