“Under Anthropic's Responsible Scaling Policy, current AI models are classified as AI Safety Level 2 (ASL-2).”