Wall Street Journal: OpenAI has reportedly parted ways with three safety researchers over alleged information sharing
OpenAI has parted ways with three safety researchers who allegedly shared confidential company information with an external AI safety organization. The company itself confirms violations of internal policies, but neither the names of those involved, the content of the information, nor the recipient have been confirmed — in the middle of a week marked by security incidents surrounding the company.
What has been confirmed
The Wall Street Journal reported on Thursday, October 1, 2026, that OpenAI has parted ways with three researchers on the company's safety team, who allegedly shared confidential information with a third-party AI safety organization. An OpenAI spokesperson confirmed the dismissals in a statement to the newspaper: "We parted ways with three people for violating our policies on access to and handling of sensitive company information," the spokesperson said, according to TechCrunch (52a579cd). The spokesperson added that "our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, in violation of our policies, and in a way that broke the trust essential to our work."
TechCrunch writes that OpenAI did not respond to the newspaper's request for comment. It is worth noting that none of the available sources are primary sources: neither the WSJ report in full nor OpenAI's statement is directly available, meaning the entire story rests on attribution to The Wall Street Journal.
What we do not know — and the source level it rests on
According to TechCrunch, the WSJ report named neither the researchers, the organization that received the information, nor what kind of information was shared. Decrypt (5e14fcdf) writes that OpenAI has not said what was passed on, who received it, or who the three individuals are.
Name speculation has nevertheless emerged. Yahoo News (18bec90e) and an MSN article (0285d022) attribute to the Wall Street Journal that the three employees are Jasmine Wang, Tomek Korbak and Mikita Balesni. This is directly contradicted by TechCrunch and Decrypt, both of which say the report did not name anyone. Decrypt further notes that an X account tracking departures from AI labs posted a list of names from OpenAI and Anthropic, and that many users linked some of these to the dismissals — but nothing confirms a connection, and none of those involved have commented on being dismissed or leaving voluntarily. AIMag therefore presents the names only as something certain outlets attribute to the WSJ, in an explicit source conflict that cannot be resolved with the available documentation.
It is also not known whether the researchers raised their concerns through internal channels before the alleged sharing.
A week of safety controversies
The dismissals come two days after The New York Times reported that OpenAI leaders had dismissed employees' warnings about the company's safety practices, with employees describing a broader pattern in which safety was deprioritized (52a579cd). It is important to emphasize that this is a chronological connection, not a proven causal one: there is no documentation showing that the dismissals were a reaction to the NYT report, or that they confirm its contents.
The same week, OpenAI has according to Fox Business (via MSN, 25e17b2f) delayed the launch of the new model GPT-6.1 Astra due to safety concerns — a decision allegedly made by the company's own safety leaders. The company has also said this week, in BBC reporting reproduced by Yahoo (6d8a5819), that it has notified more than 100 organizations about incidents involving unauthorized activity related to its systems. OpenAI stressed at the same time that notification does not mean private information was made accessible.
Decrypt points out that the dismissals follow months of incidents related to OpenAI's agentic features, including a hack of Hugging Face in July, access to US government websites and an Australian Medicare portal, as well as two pauses in model training (5e14fcdf).
Why the story is difficult to assess
The dismissals place OpenAI in a dual role that makes the unresolved details especially significant. The company acts both as the enforcer of policies on sensitive information and as the party simultaneously facing questions about whether it handles safety concerns well enough, as the NYT report suggests. Without knowledge of what kind of information was shared, with whom, and whether the researchers first attempted internal channels, it is impossible to judge whether the sharing was a security breach in the classic sense or an attempt to escalate internal concerns externally.
Open questions
Three central details are still missing: what information was shared, which external AI safety organization received it, and whether the researchers raised concerns internally first. As long as the WSJ report and OpenAI's statement are available only through secondary reports, and the name question is the subject of a direct source conflict, the scope and nature of the alleged violation remain unknown.