Anthropic consults law firms to set fair negotiation norms before AI agents start cutting deals

Why it matters

Anthropic is consulting law firms to establish fair negotiation norms for AI agents that could one day haggle over prices, handle complaints, or improve contract terms on users' behalf, according to the Financial Times. The company's own experiments show that stronger models strike better bargains, raising concerns that unequal access to capable agents could widen societal inequality. The effort builds on an existing program: Anthropic is hiring lawyers as Conceptual Reasoning Fellows to spend months testing Claude on judgment-heavy questions spanning law, philosophy, cognitive science, economics, and political science. The underlying motivation is a data problem. Fields like math and coding offer plentiful verifiable data for improving models, while the humanities lack clear-cut right and wrong answers. Anthropic therefore needs domain experts to build test cases and evaluations that strengthen Claude in these areas. Justin Curl, who researches law, institutions, and AI at the company, notes that defining quality in law is notoriously difficult, which makes building such evaluations especially challenging.