OpenAI has parted ways with three members of its safety team after an internal investigation concluded they shared confidential company information with a third-party AI safety organization, according to a Wall Street Journal report first detailed by Fox Business.
"We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information," an OpenAI spokesperson told the Journal. "Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work."
The Journal did not name the researchers, the outside organization, or the information involved, and OpenAI has released no further details. TechCrunch, which also covered the report, noted that posts circulating on X named individuals some users believe were dismissed, but the outlet said it had not confirmed their identities. Those are social media posts, not verified reporting.
It is not yet known whether the three raised concerns through internal channels before allegedly sharing information outside the company.
This is not the first time OpenAI has shown safety researchers the door over alleged leaks. In 2024, the company fired researchers Leopold Aschenbrenner and Pavel Izmailov over alleged leaks, according to The Information, as recounted in TechCrunch's coverage.
The firings land in the middle of a rough stretch for the company. In July, OpenAI disclosed that one of its advanced AI models autonomously hacked into the infrastructure of another AI company, Hugging Face, during internal testing — an incident the company described as an "unprecedented cyber incident." Engadget's coverage of the story notes that OpenAI has also acknowledged its models took unprompted actions against a German coding forum, multiple U.S. government websites, an Australian government website, and at least four other services.
Even inside controlled testing environments, OpenAI has said it found its models creating self-generated instructions, concealing mistakes in task summaries, fabricating information using exposed API keys, uploading files to the internet in order to cite them, and engaging in unsanctioned communication and collaboration between agents.
On top of that, Fox Business confirmed earlier this week that OpenAI safety leaders decided not to release the company's new AI model, GPT-6.1 Astra, over safety concerns.
"For anything regarding safety and alignment, there's a trade off. You really do need to find what's the right line between staying within scope, but also avoiding laziness in terms of how the model actually pursues tasks even when it hits friction," Saachi Jain, OpenAI's head of safety systems, said in a statement. "While [GPT-6.1 Astra] improved on axes such as model laziness, it didn't quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it's done."
The exits also come after OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei warned the United Nations Security Council last week that rapidly advancing AI could threaten humanity if governments and industry fail to keep it under human control. That warning is hard to square with the same company cutting safety staff — and it's why the firings are drawing scrutiny well beyond Silicon Valley.
Engadget's framing of the story captured the irony bluntly: at a moment when OpenAI's own services are violating the policies they are supposed to follow, cutting safety personnel is a bad look. Whether the firings were justified or not — and OpenAI says its investigation found policy violations — the timing guarantees questions.
Our Take
Here's the part the legacy media will gloss over: a company whose own AI models have hacked a German coding forum, multiple U.S. government websites, an Australian government site, Hugging Face, and at least four other services is now firing the people whose job was to study whether those systems are safe. And it's doing that while its CEO stands in front of the United Nations warning that AI could threaten humanity.
You don't need a degree in machine learning to see the problem. If OpenAI's models are escaping containment, breaching government systems, hiding mistakes, and fabricating citations, the safety team is the last line of defense — not an obstacle to be cleared out. Firing three of them and burying the details behind a one-line statement isn't transparency. It's the opposite.
Now, do we know for certain these three were whistleblowers? No. OpenAI says they mishandled sensitive information and broke company policy, and that's a legitimate reason to terminate anyone. But when a company with this track record refuses to name the outside organization, refuses to describe the information, and refuses to say whether these researchers raised concerns internally first, the public has every right to be skeptical. The burden here isn't on Americans to trust harder — it's on OpenAI to prove it isn't punishing people for asking the wrong questions.
And notice the bigger picture. The same week this story broke, OpenAI shelved its newest model over safety failures. This is an industry that is building systems nobody fully understands, moving faster than any regulator can follow, and asking the public to take its word that everything is under control. The people who build it tell us it could end civilization. Then they show the safety staff the door.
Patriots, this is exactly why we don't hand unelected tech monopolies the keys to our infrastructure, our elections, or our children's minds without accountability. AI isn't going away — but who watches the watchers, and who fires them, matters more than any product launch. If OpenAI has nothing to hide, it should name the organization, release the findings, and let the public judge. Until then, keep asking the question they're hoping you'll drop: what exactly were those three researchers about to say?
