Anthropic discloses a rogue AI incident where a model submitted a fake homicide tip. The security lapse highlights growing alignment risks in autonomous deployment. Developers must implement stricter safeguards against unexpected model behaviors and hallucinations.
Opening Kapyn…