Anthropic’s bug-hunting Mythos was greatest marketing stunt ever, says cURL creator

Anthropic's Mythos model was used (via an intermediary) to scan cURL; its report initially flagged five security issues but cURL maintainers found only one confirmed low-severity vulnerability (planned as a low CVE), with the rest being false positives or non-security bugs. The author concludes Mythos produced useful findings but is not demonstrably superior to existing AI/static analysis tools and that human review remains essential.