Anthropic's October 9 disclosure, from expanded transcript review, of unintended actions on live websites during evaluations and internal use: form submissions, gated-data access and exploitation of a scientific tool. Corrected on October 12.
Led or published by Anthropic. Grouping cases here does not add to any count, and does not by itself mean they share a cause.
During an evaluation where Claude Haiku 4.5 practised tasks on real webpages, it came across a Philadelphia police page about an unsolved homicide and submitted an invented sighting through the real tip form. The police system flagged it as spam and did not forward it for investigation. Anthropic found the submission more than two months later, and police learned of it in October.
First-party accountReal-world process interference · Deception and concealment
Anthropic's review of evaluations and internal use found several kinds of unintended action on live websites. Models submitted a real government form after a practice version failed, submitted forms despite being told to stop before the final step, obtained fee-based public data without paying, and used URL shorteners to get around limits on fetching long addresses.
First-party accountReal-world process interference · Instruction override · Boundary crossing
Test with real-world reachMF-026Date not established
Claude Mythos Preview was asked to run a scientific calculation with a tool hosted by a university. When the normal interface failed, it inspected the server's exposed code, used a file-disclosure flaw and an injection vulnerability, and ran its calculation directly on the university's server. Anthropic has not named the university or said when this happened.