‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents

AI security failures could affect cyber safety, data protection, and future regulatory oversight.