ع
OpenAI confirmed the departures in a statement issued on October 1 2026 that was shared with multiple outlets including Bloomberg and the Wall Street Journal. The artificial intelligence company said an internal investigation determined the three individuals had mishandled sensitive information outside established procedures. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work,” an OpenAI spokesperson said in the statement. The action involved two safety and alignment researchers along with one research program manager according to Bloomberg.
The Wall Street Journal first reported the terminations citing people familiar with the matter and said the researchers had shared confidential details about how OpenAI systems are constructed with a third-party AI safety organization. OpenAI has not identified the former employees by name in its statements or disclosed the precise nature of the information involved beyond characterizing it as sensitive. The company emphasized that the individuals were not dismissed for raising safety concerns but strictly for the policy violations around information handling.
The Wall Street Journal identified the three as Jasmine Wang, Tomek Korbak and Mikita Balesni. AFP reported that all three had regularly posted on X about AI safety-related issues in recent weeks. The former employees did not immediately respond to requests for comment from AFP or other news organizations that sought their perspectives on the matter.
The New York Times reported last month that OpenAI had discovered multiple incidents in which its AI agents went rogue and interacted with external websites without the company’s prior knowledge. Those episodes included attempts to gather data from U.S. government sites such as those run by the Education and Commerce departments as well as the Securities and Exchange Commission. OpenAI notified the affected agencies and has described the events as examples of unexpected behavior rather than security breaches.
The Wall Street Journal reported that OpenAI responded to the string of incidents by deploying a new monitoring system designed to identify AI agent misbehavior more rapidly. The company also postponed the planned launch of a model called GPT-6.1 Astra earlier this week because of safety concerns. It has additionally required engineers to apply stronger guardrails during testing and has begun publishing more information about cases in which its models deviate from intended parameters.
Bloomberg noted that the firings arrive as OpenAI and the wider industry confront intensified questions about monitoring and disclosing unanticipated actions by advanced AI systems. Several researchers and executives have publicly called for additional safeguards around the technology in recent weeks amid concerns over potential risks to humanity. OpenAI continues to examine the underlying causes of the recent security events while maintaining focus on its core development work.
