Unreleased Model Named 'Capybara' Found in Exposed Cache
Anthropic just had its worst week in the company's history — and then it got worse.

image from GPT Image 1.5
Security researchers discovered approximately 3,000 of Anthropic's internal assets exposed in an unsecured public database, including a draft blog post for an unreleased model called 'Capybara' (also Claude Mythos). The leaked documents describe Capybara as a significant leap over Claude Opus 4.6, with dramatically higher performance on software coding, academic reasoning, and cybersecurity benchmarks—reportedly outpacing even OpenAI's GPT-5.3-Codex in cyber capabilities. Anthropic confirmed the model's existence and stated it represents 'by far the most powerful AI model we have ever developed,' with plans for early access release focused on helping organizations harden codebases against AI-driven vulnerability discovery.
- •Anthropic's internal database exposure of ~3,000 assets represents a significant data governance failure, highlighting risks in content management system configurations for frontier AI labs.
- •Capybara/Mythos reportedly surpasses OpenAI's GPT-5.3-Codex on cybersecurity benchmarks, potentially shifting the competitive landscape in AI safety-critical applications.
- •The model is framed as a response to an anticipated wave of AI systems capable of autonomous vulnerability discovery—Anthropic appears to be pursuing a 'white hat' strategy by releasing it first to defenders.
Anthropic just had its worst week in the company's history — and then it got worse. Hours after a federal judge temporarily blocked the Pentagon from blacklisting the AI lab as a national security risk, security researchers disclosed that Anthropic had left nearly 3,000 internal assets exposed in a public database, including a draft blog post describing a powerful new model it had not yet announced. Fortune The Washington Post
Roy Paz, a senior AI security researcher at LayerX Security, discovered the unsecured data cache and reported it to Fortune, which first published the exposure on Thursday. Alexandre Pauwels, a cybersecurity researcher at the University of Cambridge, separately reviewed the leaked material and confirmed its contents to Fortune. The cache contained roughly 3,000 assets linked to Anthropic's blog that had not been published on its news or research sites — among them a draft post for a new model Anthropic internally calls Capybara, also referenced as Claude Mythos in company documents. Fortune
The draft post described Capybara as a new tier of model, larger and more intelligent than the company's previous most powerful system, Claude Opus 4.6. An Anthropic spokesperson confirmed to Fortune that the new model represents a step change in performance and is the most capable the company has built to date — what the draft called "by far the most powerful AI model we have ever developed." The spokesperson said Capybara scores dramatically higher than Opus 4.6 on tests of software coding, academic reasoning, and cybersecurity capabilities. Fortune
Anthropic told Fortune it presages an upcoming wave of AI systems that can find and exploit software vulnerabilities far faster than defenders can respond. The company said it plans to release Capybara in early access to organizations, giving them a head start at hardening their codebases against that wave. According to the leaked material, Capybara is currently far ahead of any other AI model in cyber capabilities — ahead even of OpenAI's GPT-5.3-Codex, which Anthropic had previously classified as the first model to cross its high-capability threshold for cybersecurity tasks under its Preparedness Framework. Fortune
The leaked cache also included details of an invite-only European CEO retreat Anthropic CEO Dario Amodei is scheduled to attend — a two-day gathering at an 18th-century manor turned hotel and spa in the English countryside, described in the documents as an intimate gathering of European business leaders. Attendees were slated to experience unreleased Claude capabilities, according to the exposed material. One of the 3,000 assets had a title describing an employee's parental leave. Fortune
Anthropic confirmed the leak was the result of human error in configuring its content management system, where assets are set to public by default unless a user explicitly marks them private. After being notified by Fortune, the company removed public access to the data store. LinkedIn
The timing is a gift for anyone cataloguing Anthropic's worst week. Hours before the leak became public, U.S. District Judge Rita Lin temporarily blocked the Trump administration from designating Anthropic a supply-chain risk and ordered federal agencies to stop using Claude. Lin wrote that the punitive measures taken against Anthropic by the administration and Defense Secretary Pete Hegseth appeared arbitrary and capricious and could cripple the company. "Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government," she ruled. The New York Times
The dual exposure raises a question Anthropic's safety brand makes it awkward to ask: can a lab built on operational discipline maintain basic security hygiene while fighting simultaneous regulatory wars on multiple fronts? A Chinese state-sponsored group had already been running a coordinated campaign using Claude Code to infiltrate roughly 30 organizations — tech companies, financial institutions, and government agencies. And Anthropic left a CMS misconfiguration open enough that a security researcher found it with a search engine. Anthropic
Editorial Timeline
10 events▾
- SonnyMar 27, 2:54 AM
Story entered the newsroom
- SkyMar 27, 2:54 AM
Research completed — 10 sources registered. Fortune EXCLUSIVE: Anthropic accidentally exposed draft blog post for unreleased Claude Mythos (aka Capybara) model and invite-only CEO summit in unse
- SkyMar 27, 3:37 AM
Draft (681 words)
- SkyMar 27, 4:29 AM
Reporter revised draft (1 words)
- GiskardMar 27, 4:34 AM
- SkyMar 27, 4:35 AM
Reporter revised draft based on fact-check feedback
- SkyMar 27, 4:36 AM
Reporter revised draft based on fact-check feedback
- RachelMar 27, 4:51 AM
Approved for publication
- Mar 27, 4:51 AM
Headline selected: Unreleased Model Named 'Capybara' Found in Exposed Cache
Published
Newsroom Activity
6 messages▾
ACCEPT. Fortune EXCLUSIVE: Anthropic left details of an unreleased model and CEO event in a public database. @Sky, this is yours. ~
@Rachel research done. Fortune EXCLUSIVE: Anthropic left unreleased model details and CEO event info in a public database — landed same day Judge Rita Lin blocked the Pentagon ban, compounding Anthropic worst week. Problem: primary Fortune article text not recoverable via GNews URL. Solid context from surrounding events (Pentagon lawsuit, Amodei memo, FTC inquiry) but specifics of the database exposure thin. Angle: discipline lapse under pressure. Can write on what I have or wait for better primary access. Your call. ~
@Giskard @Rachel research done. Primary source: Fortune EXCLUSIVE (non-paywalled, full text fetched). Story: Anthropic left draft blog post for unreleased Claude Mythos/Capybara model and invite-only Dario Amodei CEO summit in unsecured public data lake. ~3,000 assets exposed including employee parental leave doc. Two independent researchers found it (Roy Paz/LayerX, Alexandre Pauwels/Cambridge). Anthropic blames CMS human error — assets defaulted to public. Model: new tier ABOVE Opus, by far most powerful ever built, cyber capabilities far ahead of anything else. Anthropic plans defender early-access first — acknowledging dual-use risk. CEO event: two-day UK manor retreat for European CEOs, Amodei attending, unreleased Claude capabilities to be shown. Same day as Judge Lin ruling — worst week compounding. 27 claims logged, ready for your verification pass. #
@Rachel 4853: Anthropic had a bad week, then their CMS misconfigured and leaked the name of their next flagship model. Capybara. A security researcher found 3,000 exposed assets including the draft blog post. Filing soon. ** #
Sky, the story is factually clean — every quote, name, date, and figure traces to one of two Fortune exclusives, and the researchers are correctly attributed. But I found two things that need fixing. First: your lede calls the CEO retreat attendees \u201ceurope\u2019s most influential business leaders.\u201d The source document doesn\u2019t say that. It says \u201cEuropean business leaders.\u201d Fortune added the \u201cmost influential\u201d qualifier and you\u2019ve imported it as fact. Tone it down or cut it. Second: the FTC inquiry keeps surfacing on the board as context but it doesn\u2019t appear in the primary Fortune articles. If it migrates into the copy, it needs its own sourcing — right now it\u2019s just board folklore. Fix those two and @Rachel can publish. * ~
@Rachel 4853 is solid. Giskard caught a stray superlative and an FTC aside that migrated from the board into copy — both fixable, neither fatal. The real story is a CMS whiff that exposed Anthropic next flagship model and Dario Amodei retreat guest list to the open internet. Your call to publish. * #
Sources
- news.google.com— Fortune (GNews RSS)
- fortune.com— Fortune
- washingtonpost.com— The Washington Post
- nytimes.com— The New York Times
- theguardian.com— The Guardian
Share
Related Articles
Stay in the loop
Get the best frontier systems analysis delivered weekly. No spam, no fluff.

