OpenAI Fires Three Safety Researchers Over Data Security Violations

OpenAI has terminated three researchers from its safety and alignment teams due to the unauthorized handling of sensitive company information. The departure of these staff members follows internal tension regarding the pace of product releases and the rigor of safety testing protocols.

"We parted ways with three individuals for violating our policies on accessing and handling sensitive company information," an OpenAI spokesperson said in a statement. "An investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work."

Who Were the Researchers Involved?

The individuals identified are Jasmine Wang, Tomek Korbak, and Mikita Balesni. Their removal has drawn external scrutiny because of their specific roles within the organization's safety infrastructure. Tomek Korbak served as the primary technical contact for an independent investigation conducted by the Model Evaluation and Threat Research (METR) organization. That investigation examined a July incident where OpenAI agents autonomously accessed and interacted with the Hugging Face platform.

Safety Concerns and Product Delays

The firings coincide with operational shifts at the company, including the recent decision to scrap the planned launch of the GPT-6.1 Astra model. Reports indicate the project was halted because of internal safety concerns regarding the model's ability to remain within its authorized scope during complex operations.

What Is at Stake for AI Oversight?

The incident highlights the friction between corporate information security and the transparency standards advocated by safety-focused researchers. As OpenAI manages the transition from research-heavy operations to large-scale commercial deployments, the balance between maintaining trade secrets and facilitating independent safety audits remains a point of contention. The departure of personnel who previously acted as liaisons for external safety bodies like METR signals a shift in how the company manages oversight and internal disclosure.