ai-safetyrisk-assessmentresearch The Difference Between Very Low and Low Anthropic's decision to raise its catastrophic AI risk estimate reflects a critical moment of reckoning for the future of building intelligent systems.