In September 2026, Anthropic committed to allowing independent third-party assessments of its artificial intelligence models and safety practices