Review of LLM security challenges, covering external threats like adversarial attacks and internal AI safety risks from model misbehavior. It highlights MLSecOps, AI as emerging threat actors, and knowledge editing for subtle supply chain attacks.