Anthropic2026-04-14
Claude Mythos Preview Launches with Advanced Cybersecurity Capabilities via Project Glasswing

AI Analysis
Anthropic released Claude Mythos Preview to a vetted group of roughly 50 companies through Project Glasswing, featuring advanced cybersecurity capabilities powered by strong coding and reasoning skills. During internal testing, Mythos autonomously discovered and exploited zero-day vulnerabilities, prompting restrictions due to its potent cyber-attacking abilities. Reuters reports the model could supercharge complex cyberattacks, posing significant risks to the banking industry.