Project Glasswing is using advanced AI models to expose hidden vulnerabilities across critical open-source software, putting new pressure on already overwhelmed maintainers.

Anthropic's Claude Mythos Preview found 23,000 potential vulnerabilities in 1,000 open source projects, with over 1,000 confirmed as critical severity flaws.

Anthropic restricts access to its Mythos AI model after it autonomously found thousands of software vulnerabilities, rattling crypto and cybersecurity markets.