How Anthropic's Mythos Model Has Already Uncovered Over 10,000 Vulnerabilities The latest report from Anthropic on its Project Glasswing initiative has shed light on the staggering capabilities of Claude Mythos Preview.
In just a month, this model has discovered over 10,000 vulnerabilities.
While this achievement may seem remarkable, it also raises fundamental questions about the ethics of using AI to uncover security flaws and the potential risks associated with unleashing such powerful technology.