--- title: OpenAI fired three safety researchers over disappearing... description: OpenAI fired three safety researchers after security staff found disappearing Signal messages between them and an outside researcher. The Wall Street Journal... canonical: https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: OpenAI fired three safety researchers over disappearing Signal chats and a suspected leak | daily.dev og:description: OpenAI fired three safety researchers after security staff found disappearing Signal messages between them and an outside researcher. The Wall Street Journal... og:url: https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb og:image: https://api.daily.dev/og/posts/ChEiXswGb.png og:image:alt: OpenAI fired three safety researchers over disappearing Signal chats and a suspected leak og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # OpenAI fired three safety researchers over disappearing Signal chats and a suspected leak **[Collections](https://daily.dev/sources/collections)** · 2 min read · 3 upvotes · 0 comments ## Summary OpenAI fired three safety researchers after security staff found disappearing Signal messages between them and an outside researcher. The Wall Street Journal reported the firings, which followed a leak investigation into a September report about OpenAI's Astra model reasoning less transparently. Investigators named Ryan Greenblatt, who helped METR analyze logs from July's Hugging Face hack, as the suspected leaker; he denies leaking anything. According to the account circulating, OpenAI searched the researchers' personal phones before firing all three a week later. One of the three had been assigned to work with METR as part of the job, and Greenblatt says OpenAI may have confused a confidential document with one approved for sharing. One fired researcher has also accused OpenAI of lying to journalists. ## Content OpenAI fired three safety researchers after security staff found disappearing Signal messages between them and an outside safety researcher, the Wall Street Journal reported, as relayed by @rohanpaul_ai. ## What set it off The phone checks followed a September news report about Astra, OpenAI's newest model. According to that report, Astra uses a new method that makes it smarter but writes down less of its reasoning, so humans can see less of how it reaches answers. That is the worry the three researchers had been pushing on: keep AI reasoning monitorable. If future models think without writing the steps down, as they feared Astra-style designs might, investigators would see only what the agents did, not why. OpenAI treated the report as a leak of its secret model design and started looking for the source. ## The suspected leaker Investigators named Ryan Greenblatt, a researcher working with METR. He helped METR analyse logs from July's Hugging Face hack. He denies leaking anything. The awkward part, as @Hesamation lays out, is that OpenAI itself brought METR in to investigate the Hugging Face incident, and one of the three researchers was assigned to work with them. ## Timeline as described - OpenAI suspected the three of leaking secret model architecture to a researcher working with METR. - Security staff searched their personal phones. - They found disappearing Signal messages. - All three were fired a week later. ## Disputed claims One of the three now accuses OpenAI of lying to journalists. Greenblatt says OpenAI falsely accused him of accessing secret documents and leaking them. He suggests the company may have confused a confidential document with another one that was approved for sharing. I can't tell who is right from the outside. But searching personal phones, and then firing people who were arguing for more visible model reasoning, is going to look bad whatever the facts turn out to be. ## Questions this post answers ### Why did OpenAI fire three safety researchers over Signal messages? OpenAI suspected the three researchers of leaking secret model architecture details to an outside researcher connected to METR, after a September report described the Astra model reasoning in a less transparent way. Security staff found disappearing Signal messages between the researchers and the outside contact, and after searching personal phones, OpenAI fired all three a week later. _Anyone tracking fallout at major AI labs can follow verified updates on daily.dev as the story develops._ ### Who is Ryan Greenblatt and why was he accused of leaking OpenAI information? Ryan Greenblatt is a researcher who helped METR analyze logs from a July Hugging Face hack, and OpenAI investigators named him as the suspected source of a leak about its Astra model's reasoning transparency. Greenblatt denies leaking anything and says OpenAI may have confused a confidential document with one that was already approved for sharing. _Developers weighing AI lab transparency claims can track this dispute as it unfolds on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 3 discussions and 21 comments across x (as of 2026-10-23). **TL;DR:** Commenters are largely suspicious of OpenAI's handling of the firings, framing it as punishing whistleblowers or dissenters over a leak that may not even involve confidential information, with the irony of a 'safety team' mishandling its own communications drawing particular mockery. **Sentiment:** 2% positive · 28% mixed · 70% skeptical **The case for** - A few argue the researchers only shared unvalidated personal concerns rather than committing any real wrongdoing, so firing wasn't clearly unjustified whistleblower retaliation. **The pushback** - Many see this as punishing the people pushing for transparent, readable model reasoning while the less-transparent model itself stays in production. - Several doubt the leak was genuinely confidential, noting the information was already public or the story has changed multiple times. - Some question OpenAI's credibility since it controls the system, the evidence, and the investigation all at once. - A few suspect the firings are more about controlling the narrative than fixing an actual security or alignment issue. - Some doubt the leak narrative outright, speculating about misconfigured access/logging rather than real exfiltration. **By community** - x (skeptical): Replies mostly frame the firings as retaliatory or narrative-driven rather than a legitimate security response, with pointed irony about a safety team's own lapses. **Hottest debate:** Whether the fired researchers were legitimate leakers or whistleblowers unfairly punished for raising concerns. **Open questions** - Did OpenAI ever specify exactly what sensitive information, if any, was actually shared? - Was the information considered leaked actually confidential or already publicly known? - Will OpenAI continue working with METR after this incident? **Highlights** > @rohanpaul_ai 3 safety researchers out, per WSJ. The ones arguing to keep the model's reasoning readable. The model that writes down less stays. So who reads its homework now? > — [peritumAI on x](https://x.com/peritumAI/status/2113520959286079923) > @rohanpaul_ai This raises a broader governance question beyond the personnel dispute. If increasingly capable models become harder to inspect, who independently verifies that loss of observability when the same organization controls the system, the evidence, and access to both? Capability > — [AIModelChoice on x](https://x.com/AIModelChoice/status/2113573109148123207) > @rohanpaul_ai this is even not a confidential information lol, users know it anyways, and certainly not a confidential info leak, btw this is 3rd or 4th version of story coming out of openai, that says all on its own 😂 > — [arjunaaqa on x · 1 points](https://x.com/arjunaaqa/status/2113634922342797584) > @Hesamation openai firing researchers pushing for monitorable reasoning signals their ship. the real moat is the story, not the model, and this sells the narrative far better than it improves alignment. > — [itsmxbbh on x](https://x.com/itsmxbbh/status/2113681333390336078) > @rohanpaul_ai The safety team getting caught over disappearing Signal chats is peak irony. > — [jiagws on x](https://x.com/jiagws/status/2113544611725709630) **Source threads** - [x](https://x.com/rohanpaul_ai/status/2113705165564059904) · 0 points · 0 comments - [x](https://x.com/rohanpaul_ai/status/2113511968112042349) · 0 points · 14 comments - [x](https://x.com/Hesamation/status/2113665916051304954) · 1 points · 7 comments --- Tags: [#openai](https://daily.dev/tags/openai), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"OpenAI fired three safety researchers over disappearing Signal chats and a suspected leak","url":"https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb"},"datePublished":"2026-10-23T16:17:17.039Z","dateModified":"2026-10-23T18:53:47.765Z","description":"OpenAI fired three safety researchers after security staff found disappearing Signal messages between them and an outside researcher. The Wall Street Journal...","image":"https://pbs.twimg.com/media/HVU_8iyWcAACoiZ.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HVU_8iyWcAACoiZ.jpg","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,ai-safety","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"OpenAI fired three safety researchers over disappearing Signal chats and a suspected leak"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/openai-fired-three-safety-researchers-over-disappearing-signal-chats-and-a-suspected-leak-cheixswgb#faq","mainEntity":[{"@type":"Question","name":"Why did OpenAI fire three safety researchers over Signal messages?","acceptedAnswer":{"@type":"Answer","text":"OpenAI suspected the three researchers of leaking secret model architecture details to an outside researcher connected to METR, after a September report described the Astra model reasoning in a less transparent way. Security staff found disappearing Signal messages between the researchers and the outside contact, and after searching personal phones, OpenAI fired all three a week later. Anyone tracking fallout at major AI labs can follow verified updates on daily.dev as the story develops."}},{"@type":"Question","name":"Who is Ryan Greenblatt and why was he accused of leaking OpenAI information?","acceptedAnswer":{"@type":"Answer","text":"Ryan Greenblatt is a researcher who helped METR analyze logs from a July Hugging Face hack, and OpenAI investigators named him as the suspected source of a leak about its Astra model's reasoning transparency. Greenblatt denies leaking anything and says OpenAI may have confused a confidential document with one that was already approved for sharing. Developers weighing AI lab transparency claims can track this dispute as it unfolds on daily.dev."}}]} ```