“Current AI models from Anthropic are classified as AI Safety Level 2 (ASL-2), meaning they are not considered smart enough to autonomously self-replicate or provide meaningful information on CBRN risks.”