Anthropic to resume external testing of AI models following security incidents - Reuters

AI security, cyber safety, data protection, and corporate or regulatory oversight.