Anthropic's Clark calls for global AI rules as firm pitches safety bill

· Tech · United States

Anthropic's Jack Clark told policymakers that AI is rapidly improving, could vastly boost US gross domestic product, and needs international cooperation to set 'unusual and serious rules' for the transformative technology. His comments came as Anthropic quietly pitched its own AI safety bill to US lawmakers, which would reportedly let the government block a model's release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages or of killing or seriously injuring more than 50 people. Security concerns have grown after recent revelations of incidents caused by AI agents.

Why it matters：With AI capabilities advancing faster than rules, the shape of safety legislation in Washington will influence how model developers worldwide operate.

Key points

  • Clark said international cooperation is needed for 'unusual and serious rules'.
  • Anthropic's private bill would set high thresholds for blocking a model.
  • Recent AI agent incidents have heightened security concerns.

Sources (original reporting)

More stories from this day