Published: October 8, 2026 · Last updated: October 8, 2026
TL;DR: The three researchers OpenAI fired on October 1 over an unexplained information leak published an open letter Thursday disputing that account, according to TechCrunch and The Information. They say they were pushed out for prioritizing safety over OpenAI’s near-term commercial interests, are asking the board to keep the company’s models auditable rather than move to harder-to-monitor “opaque-reasoning” systems, and warn the firings will make other employees think twice before raising concerns.
OpenAI fired three researchers at the start of October over what it called the mishandling of sensitive information. It never said what was shared, who received it, or why it was serious enough to end three careers on the spot. On Thursday, the three researchers made clear they aren’t going to let that vague explanation stand as the only version of events.
The firings first surfaced on October 1 in a Wall Street Journal report, which said an internal investigation found the researchers had shared sensitive company information with an outside AI safety evaluation group. OpenAI didn’t name the group, specify what was shared, or explain what harm it supposedly posed, a gap we covered in detail when the story broke. That silence is a big part of what the researchers are now pushing back on.
TechCrunch and The Information both reported Thursday that the three published an open letter disputing OpenAI’s account entirely. According to that reporting, the letter says they were terminated for prioritizing safety over OpenAI’s near-term commercial interests, not for leaking anything that put the company or the public at risk. They’re also warning that firing people this way, with no public accounting of the supposed wrongdoing, sends an unmistakable message to everyone still inside OpenAI: raise a safety concern the wrong way and you could be gone before you ever learn why.
The letter reportedly goes further than defending their own conduct. The three are asking OpenAI’s board to commit to keeping its models’ reasoning auditable rather than moving toward what’s being described as opaque-reasoning systems, and they’re calling for independent safety audits of the company’s frontier models. Chain-of-thought monitoring, the practice of letting safety researchers read a model’s step-by-step reasoning before it produces a final answer, is one of the few tools anyone has for catching a model’s intent before it acts on it. If a lab shifts to architectures where that reasoning gets compressed or hidden for performance reasons, that tool disappears, and nobody inside or outside the company gets a reliable way to check what the model is actually planning.
This isn’t the first time a group of departed OpenAI employees has aired safety concerns publicly after leaving on rough terms. A similar wave happened in 2024, when former staff including Daniel Kokotajlo signed onto a “right to warn” letter arguing that restrictive exit paperwork was keeping people who’d seen safety problems up close from saying anything about them. That episode ended with OpenAI publicly loosening some of those exit terms. Whether this one follows the same arc depends entirely on how OpenAI responds, and as of Thursday the company hadn’t issued a public statement addressing the letter’s specific claims.
The part of this story worth actually worrying about isn’t who’s telling the truth about one set of firings. It’s the pattern underneath it. Every frontier lab is racing to ship more capable models faster than the last one, and the people whose entire job is to slow that down and ask whether it’s safe keep ending up on the outside, whether through resignation, restructuring, or in this case, termination. A chilling effect doesn’t announce itself when it happens. It shows up months later as the question nobody asked in a meeting, because the last person who asked it isn’t on the payroll anymore.
Related: This follows our earlier coverage of OpenAI firing three safety researchers over a leak it still hasn’t explained, the original story this letter responds to, and it’s worth reading alongside NYC’s first-ever AI safety hearing for a sense of how little public accountability AI labs face even when lawmakers are the ones asking the questions.
Bottom Line: An open letter from three fired employees doesn’t settle anything by itself. People who get pushed out always have a story to tell, and this one is still built on reporting about a letter, not the letter’s full text. But OpenAI had a chance to close this out on October 1 with a real explanation and gave a one-line statement instead, and that’s exactly the kind of vacuum that lets the fired researchers’ version become the most detailed account in the room. If OpenAI wants its side to matter here, it needs to actually say something.
Follow Teck Hustlers: Facebook · X · Instagram · LinkedIn · TikTok · Pinterest
