OpenAI Parts Ways With Three Safety Researchers Over Information-Handling Violations
OpenAI has parted ways with three researchers from its safety team after an internal investigation found violations involving the handling of sensitive information, while separate reporting alleged that confidential material was shared with an outside group.
The company confirmed the departures but provided limited public detail about the conduct involved. An OpenAI spokesperson told CBS News that an internal investigation found misconduct outside approved procedures and said the behavior violated policies governing access to sensitive information.
The Wall Street Journal reported on October 1, citing people familiar with the matter, that the researchers allegedly shared confidential information externally. OpenAI has not publicly identified the researchers, disclosed the information in question or named the organization that allegedly received it.
OpenAI Investigation Centers on Confidential Information
OpenAI said the investigation found a pattern of misconduct involving confidential research accessed by individuals working on the company's safety efforts.
The company has emphasized that safety researchers require broad access to internal research and systems as part of their work. That access, however, also places heightened importance on procedures governing the use and handling of sensitive material.
OpenAI's public statement did not specify what information was allegedly transferred or explain the circumstances surrounding the reported disclosure. As a result, details about the alleged external sharing remain unresolved.
The publicly available account therefore consists of two related but distinct elements: OpenAI's confirmation that three employees violated internal information-handling policies and separate reporting that they allegedly shared confidential material with an external organization.
OpenAI has not publicly identified that organization.
Researcher Departures Come Amid Broader OpenAI Safety Scrutiny
The departures come as OpenAI faces wider attention over the safety of increasingly capable artificial intelligence systems.
The company recently withheld GPT-6.1 Astra from release after internal testing identified shortcomings against its safety and alignment standards. OpenAI confirmed the decision on September 28, ahead of the model's planned October debut, according to Reuters.
Safety systems head Saachi Jain cited concerns related to authorization, permitted activities and the way the model reported its actions. Those issues concerned the model's behavior during testing and were separate from the information-handling investigation involving the three researchers.
OpenAI has also experienced earlier departures involving members of its safety leadership. Co-founder Ilya Sutskever and researcher Jan Leike left the company in 2024, after which OpenAI dissolved the superalignment team they had led.
The circumstances differ from the latest departures. OpenAI has attributed the three researchers' exits to violations involving confidential information rather than disagreements over the company's safety strategy.
The company has also not publicly linked the internal investigation into the researchers with its decision to delay GPT-6.1 Astra. Without further disclosure, the specific nature of the alleged information sharing and its relationship to the researchers' work remain unclear.
Writer: Marcus RenfieldCrypto Market Analyst & Onchain WriterMarcus Renfield covers cryptocurrency markets with a focus on onchain data, Bitcoin price action, and emerging market narratives. His writing examines how capital flows, network activity, and broader market structure influence short- and medium-term trends.He aims to provide clear, data-informed analysis for readers seeking a deeper understanding of crypto market dynamics.