Anthropic said it would not accede to the Pentagon's request to eliminate safeguards from its AI systems, despite threats to deem the company a 'supply chain risk' and remove it from the Department of Defense's systems