In Brief: OpenAI parted ways with three safety researchers following an internal investigation into alleged breaches of company policies governing confidential information. Neither OpenAI nor
OpenAI has cited breaches of its information-handling policies in the departure of three researchers from its safety team. According to The Wall Street Journal’s October 1 report, sources familiar with the matter alleged researchers shared confidential information externally.
However, OpenAI’s public confirmation provided fewer details than the account describing the alleged sharing of information with an external group. Its statement confirmed three departures and attributed the decision to violations involving access to and handling of sensitive information.
An OpenAI spokesperson told CBS News that an internal investigation established misconduct outside the company’s approved procedures. Moreover, the spokesperson described the conduct as a breach of policies and the trust necessary for internal work.
OpenAI also reported a pattern of misconduct involving the handling of confidential research by individuals with access to that material. Its explanation emphasized the extensive internal access safety teams require when examining the company’s research and systems.
Despite those findings, OpenAI has not publicly named the researchers or explained precisely what information they allegedly shared externally. Neither the company nor the cited reports identified the outside organization that allegedly received the confidential material.
Those omissions leave several central details unresolved, including the nature of the information and the circumstances surrounding its alleged transfer. The public account therefore combines OpenAI’s findings about policy violations with separately reported allegations concerning an outside recipient.
Also Read: XRP and Zcash Lead Major Coins as Neon Surges 282.7%
Meanwhile, the departures come as OpenAI faces scrutiny over the behavior of increasingly autonomous artificial intelligence systems. The company recently withheld GPT-6.1 Astra from release because internal testing found shortcomings against its safety and alignment standards. Reuters reported that OpenAI confirmed the decision on September 28, ahead of the model’s planned October debut.
Additionally, safety systems head Saachi Jain identified concerns involving authorization, permitted activities, and how the model reported its actions. Those concerns addressed the system’s behavior during testing, while the researcher departures involved alleged misconduct by employees.
OpenAI has also experienced prominent departures from its safety leadership, although the circumstances surrounding those exits differed from this case. Co-founder Ilya Sutskever and researcher Jan Leike left in 2024, and OpenAI subsequently dissolved the superalignment team they had led.
By contrast, OpenAI attributes the three researchers’ departures to information handling violations rather than disagreements over its approach to safety. The company has not publicly connected its internal confidentiality investigation with the decision to withhold GPT-6.1 Astra from release. OpenAI’s stated explanation remains an internal policy breach, while undisclosed details limit public understanding of the alleged information sharing.
Also Read: Hyperliquid Policy Center Urges EU to Clarify Rules for Onchain Derivatives
The post OpenAI Parts Ways With Three Safety Researchers Over Alleged Confidentiality Breaches appeared first on 36Crypto.
