Overview
- Anthropic has hosted six METR evaluators on site with company hardware, office space and near‑employee access to internal tools so the nonprofit can test model risks and publish findings.
- The company proposed a Public Benefit Fund that would start with a 5% share stake and can grow as valuation milestones are met while it plans to list under a Public Benefit Corporation structure.
- Reported financing efforts include roughly $60 billion of debt to buy chips with parts of the package carrying yields at or above 10%, a cost that reflects higher investor risk premiums for AI infrastructure.
- Anthropic is consulting law firms on rules for AI agents that would bargain or act for users to define delegation limits, liability and when human sign‑offs are required.
- Internal data and public comments together show fewer junior technical hires and an internal economic model projecting large GDP gains from AI by 2030, raising questions about job shifts and the need for new policy responses.