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