Anthropic's magic code-sniffer: More Swiss cheese than cheddar, for now

**Executive summary:** This opinion article assesses Anthropic's Mythos vulnerability-finding tool and argues that while such AI tools will increasingly automate detection of known classes of bugs and change the security landscape, the piece contains no technical incident details, indicators, or reports of active exploitation—it is a forward-looking commentary on benefits, risks, and the need for human expertise.