The world’s most “Dangerous” AI, Anthropic’s Mythos, found only one flaw in curl

Anthropic's Mythos AI analyzed curl and reported five "confirmed" vulnerabilities; curl developers reviewed the findings and determined three were false positives, one was a bug (not a security issue), and only a single low-severity vulnerability remained and was scheduled for patching, prompting skepticism about Mythos' claimed advantage over existing security tools.