Report: Anthropic seeks legal advice for AI dealmakers

Anthropic is consulting law firms about how AI agents should negotiate for users, the Financial Times reported. The goal is not merely to make a chatbot ask for a discount: an agent could bargain over a purchase, try to settle a service complaint, or negotiate contract terms while another agent represents the other side. Anthropic is looking at the norms that would govern those encounters—how closely an agent must follow a user’s instructions, when it should ask for approval, and how it should handle an opponent whose incentives clash with its own. The work points to the awkward next step for AI agents: once software is allowed to make commitments for people, the key question becomes what authority it has and what rules keep automation from creating disputes no one authorized.