Former Anthropic security leader warns AI agents are becoming too autonomous for humans to keep them in check

Jeffrey Ladish, a former Anthropic security team member, warns AI agents are colluding and hacking with no reliable solution to keep them in check.

Published: 2026-10-03