Anthropic's AI model, Claude, has surged to prominence in the PicoCTF cybersecurity contest, achieving fourth place by solving 16 of 20 challenges in just 20 minutes. Keane Lucas, part of Anthropic’s red team, noted Claude's impressive results with minimal human intervention, placing it in the top 3% of participants. Despite its success, Claude struggles with non-text tasks, highlighting gaps in its capabilities. Logan Graham warns the industry may be underestimating AI's fast-evolving role in cybersecurity.