Three safety researchers who once worked on the inside of OpenAI, one of the most closely watched builders of artificial intelligence, have now spoken for the first time since their dismissals, and their account turns on a missing piece: they say no one ever told them what they were alleged to have leaked. The case grew out of the disappearing messages they used on the encrypted app Signal, a habit common among people who discuss sensitive work in private, which became the thread by which their jobs were pulled apart. It is a small human episode inside a larger, older story about who controls
Fired OpenAI Safety Researchers Speak Out for First Time, Disputing the Reasons for Their Dismissal
They say no one ever told them what they leaked.
Mark
Why is this one getting so much attention? Firings happen at tech companies all the time.
Mimi
Because these are safety researchers at OpenAI, and because the account they've now given is that they were never told what they supposedly leaked. That's an unusual place to stand.
Mark
What actually triggered it, as far as reported?
Mimi
The coverage points to disappearing messages on Signal. Reports also say the company asked them to unlock their phones before firing them.
Luke
Careful there. Those details come from outlets summarizing the case, and what we have is mostly headlines. I can't tell you from this material who exactly asked for what, or when.
Mark
And the claim that they were never told what was leaked?
Luke
That's the researchers' statement. It's their first-person account, not a confirmed finding. OpenAI's side of it isn't in what we have.
Mimi
Which is exactly why the interview matters. It puts their version on the record, and it invites a response.
Mark
Does using disappearing messages say anything on its own?
Luke
Not by itself. It's a common practice, and nothing here establishes that it proves a leak occurred.
Mimi
The open question is the content. Until someone says what was allegedly disclosed, the story has a hole in the middle.
Mark
So what do we watch?
Luke
Whether the company answers specifically, and whether anyone shows what the alleged leak was.
Le Pouls
- Three OpenAI safety researchers were dismissed, and the reasons traced back to disappearing messages on Signal, a detail that turned private habit into public suspicion.
- Reports say the company asked the researchers to unlock their phones before firing them, a demand that raised questions about how far an employer's inquiry should reach.
- The fired researchers insist they were never told what they allegedly leaked, leaving the central accusation unanswered and the dispute stuck on the question of specifics.
- Their first interview, released through NPR's Machine Gods series, has drawn coverage from several outlets, keeping pressure on OpenAI to explain itself in detail.
- The story is now heading toward a test of whether the company will clarify its leak claims, and whether the case reshapes how AI labs treat internal communications and safety staff.
Three safety researchers who once worked on the inside of OpenAI, one of the most closely watched builders of artificial intelligence, have now spoken for the first time since their dismissals, and their account turns on a missing piece: they say no one ever told them what they were alleged to have leaked. The case grew out of the disappearing messages they used on the encrypted app Signal, a habit common among people who discuss sensitive work in private, which became the thread by which their jobs were pulled apart. It is a small human episode inside a larger, older story about who controls information when the stakes of a technology feel very high, and the answer to what was actually disclosed has not yet been given.
Three OpenAI safety researchers who were fired over alleged leaks have finally spoken publicly, and their central claim is strikingly simple: no one ever told them exactly what they were supposed to have leaked. Their first interview, released by NPR as part of its Machine Gods series, follows a run of coverage that kept returning to the same details: disappearing messages on the encrypted app Signal, a reported request that the researchers unlock their phones, and then their dismissal.
The gap in the account is the heart of the matter. The company's stated reasons center on messaging practices and alleged leaks, while the researchers dispute those reasons and say they were never shown what material was supposedly disclosed. The reporting available does not resolve what the alleged leak was.
The case also reflects a familiar tension at AI labs, where safety researchers sit close to sensitive questions about how powerful systems are built, and where private messaging is common. Here, ordinary habits became grounds for suspicion. What comes next is whether OpenAI answers the researchers' account in detail and clarifies what it considers leaked, and whether the interview sharpens scrutiny of leak investigations and protections for those who raise safety concerns.
Three OpenAI safety researchers who were fired over alleged leaks have given their first interview since the dismissals, and the central point of their account is simple: they say no one ever told them what, specifically, they were supposed to have leaked.
The interview was released by NPR as part of its Machine Gods series. It follows a run of coverage from several outlets, each describing the case from a different angle. The Wall Street Journal's headline framed it around the Signal chats that cost the researchers their jobs. The San Francisco Standard reported that OpenAI fired researchers for using disappearing messages on Signal. FourWeekMBA reported that the company asked three researchers to unlock their phones and then fired them. The Decoder summarized the researchers' own position: they were never told what they allegedly leaked.
Put together, the reporting sketches a sequence. Three safety researchers at OpenAI were accused in connection with leaks. The use of disappearing messages on the encrypted app Signal became a focal point. According to reports, the company asked the researchers to unlock their phones. Then they were dismissed. The interview is the first time the three have spoken publicly about what happened.
The gap they describe is the sharpest part of the story. The company's stated rationale, as reported, centers on the messaging practices and the alleged leaks. The researchers dispute the reasons given for their dismissal and say they were not shown what material was supposedly disclosed. That leaves an open question about what the alleged leak actually was, and the available reporting does not answer it.
The case also sits inside a broader, familiar tension at AI labs. Safety researchers work close to the most sensitive internal questions about how powerful systems are built and tested, and companies have an interest in controlling that information. Employees, for their part, often use private channels to talk with colleagues. Disappearing messages on Signal are a common feature of that kind of communication, yet in this case they became grounds for suspicion.
The source material here is thin on specifics. It consists of headlines and a short summary rather than a full transcript, so details such as the researchers' names, the dates of the firings, and OpenAI's formal response are not part of what can be laid out here. What is established is the outline: three safety researchers, Signal messages set to disappear, a request to unlock phones, dismissals, and the researchers' assertion that the specifics of the accusation were never put to them.
What to watch next is whether OpenAI responds in detail to the researchers' account, and whether it clarifies what it considers to have been leaked. The interview may also sharpen scrutiny of how AI labs run leak investigations, handle internal communications, and treat the people tasked with raising safety concerns.
Citations marquantes
They were never told what they allegedly leaked.— The fired researchers, as summarized by The Decoder
Envie de l'histoire complète ?
Lire l'original sur Google News ↗
Technology