Anthropic wants Congress to mandate independent safety tests for powerful AI models

Anthropic is urging Congress to require independent safety evaluations for AI models trained above a specific compute threshold. The company also seeks government authority to prevent the deployment of models deemed dangerous.