Anthropic Mythos draws hype backlash

Claude Mythos Preview is Anthropic’s tightly restricted cybersecurity model, unveiled on April 7, 2026 through Project Glasswing as a system that can autonomously find and chain serious software exploits. This Reddit thread reacts to fresh skepticism around those claims, arguing Anthropic has generated more fear marketing than publicly verifiable proof because most of its strongest exploit evidence remains withheld under responsible-disclosure constraints.