Safety guardrails loosen as AI rivalries grows

As large language models grow more powerful and less predictable, AI companies are loosening safety guardrails in the race to be first—a shift that some warn could lead to catastrophe. Anthropic, long viewed as the most safety-focused major AI lab, recently revised a key safeguard—narrowing the conditions under which it would delay developing or releasing a model that could pose catastrophic risk. Anthropic's recalibration comes amid a dispute with the Trump administration. The company refused to allow its models to be used for autonomous weapons or domestic surveillance. The Defense Department responded by cutting use of Claude and labeling the firm a supply chain risk. That highlights another problem with competition. Even if one company refuses on safety grounds, another is likely to step in. Hours later, OpenAI announced a deal to provide models for classified networks.