It's so cool that anthropic is setting up a double-sided protection racket where it will profit from the massive token burn of attackers and defenders with a tool specifically designed to generate exploits and their only observable mitigation is a clientside system prompt that sternly warns the LLM to be good and not do malware
https://red.anthropic.com/2026/mythos-preview/
Claude Mythos Preview \ red.anthropic.com

@jonny Sigh. They quote Mickens as from a totally serious, authoritative source, when what he writes is (delightful) satire. They use "hallucinations" as if talking of a completely genuine, inevitable quirk of their system, preventable by the use of an actually deterministic, focused tool like memory sanitizers. They repurpose "Given enough eyeballs, all bugs are shallow" (an already-disproven quip) to equating their tool to intentional eyeballs.