Anthropic is consulting law firms on fair “negotiation norms” for AI agents that bargain for their users, the Financial Times reported

Anthropic is seeking advice from law firms on fair “negotiation norms” for AI agents, the Financial Times reported on Sunday 18 October, citing two people familiar with the company's thinking, according to PYMNTS and The Decoder. The agents in question would haggle over prices, settle complaints or seek better contract terms for the people they act for.

PYMNTS's account of the FT report lists the open questions: how much human input the agents should seek, and how suspicious they should be of their AI counterparts without becoming “paranoid” or “conspiratorial”. No law firms are named, and Anthropic declined to comment to the FT.

The Decoder quotes Justin Curl, of Anthropic's technical staff, who works on law, institutions and AI, saying that in law it is “notoriously hard to define quality”. The page does not make clear whether that line comes from the FT or from The Decoder's own reporting. The Decoder ties the work to Anthropic's hiring of “Conceptual Reasoning Fellows”.

The FT's own page is behind a paywall and was not opened, so the report rests on these two open accounts of it. It points to agents that deal with other agents on a user's behalf, where the questions are less about what a model can do than about how hard it should push and when it should ask.

• Reported Anthropic is consulting law firms on negotiation norms for AI agents, the Financial Times reported, according to PYMNTS. PYMNTS
• Reported The questions include how suspicious agents should be of their AI counterparts without becoming “paranoid” or “conspiratorial”, PYMNTS reported, citing the FT. PYMNTS
• Reported The Decoder, citing the Financial Times, reported that Anthropic is consulting law firms on fair negotiation norms before its agents start cutting deals. The Decoder