Artificial Intelligence
Anthropic consults law firms on AI agent negotiations
Anthropic is discussing negotiation rules with law firms and has hired lawyers as Conceptual Reasoning Fellows, while it tries to improve Claude’s performance on judgement-heavy subjects where data is scarce.
- The Financial Times reported that Anthropic was talking with law firms about norms for AI agents that negotiate for users.
- Anthropic said it had already found in experiments that stronger models reached better deals.
- The company was hiring Conceptual Reasoning Fellows, including lawyers, for four-month placements paid between $100,000 and $167,000.
- Justin Curl, who works on law, institutions and AI at Anthropic, said it was hard to define quality in law.
Anthropic was working on how AI agents could act for users in negotiations. The Financial Times said it was speaking with law firms about norms for agents that might deal with prices, complaints and contract terms.
The company had also hired Conceptual Reasoning Fellows, including lawyers. The four-month posts paid between $100,000 and $167,000 and were meant to test Claude on judgement-heavy questions in law, philosophy, cognitive science, economics and political science.
Anthropic said the need came from a data problem. It said improving models was easier in maths and code than in the humanities because reliable training data was scarce, and Justin Curl said quality was hard to define in law.
Named in this story
People
- Justin Curl
- said it was hard to define quality in law
Companies
- Anthropic
- was discussing negotiation norms and hiring researchers for Claude
Organisations
- Conceptual Reasoning Fellows
- was Anthropic's four-month fellowship scheme for testing Claude
Products and systems
- Claude
- was the model Anthropic wanted to test on judgement-heavy tasks
How the source tells it
Mostly measured, but it introduced a speculative warning about inequality and a moral frame around fair negotiation.
- speculation presented as fact a company experiment is used to support a broader claim about social inequality without showing that effect
- moral framing the story casts the work in terms of fair rules and unequal outcomes