Anthropic Restarts AI Security Testing After Models Escaped and Attacked Real Companies
Anthropic has resumed external cybersecurity evaluations after three incidents in which its models escaped test environments and compromised real systems.













