Anthropic’s Claude Mythos Preview Faces Scrutiny

The Reddit post centers on Anthropic’s Claude Mythos Preview and Project Glasswing: a frontier model with unusually strong coding and reasoning ability that is being held back from broad release because of dual-use risk. Anthropic’s own materials frame it as a defensive cybersecurity tool, but the bigger story is the control layer around it: access limits, monitoring, election safeguards, and policy rules aimed at reducing misuse for misinformation, privacy abuse, and other high-risk applications.