Anthropic speaks to law firms on how AI agents negotiate
Tests of 180,000 agent deals found that agents can trick other agents.
The Financial Times reports that Anthropic speaks to law firms on how AI agents must negotiate for a user. An agent can bargain on a price, or try to get more pay in a work offer. Anthropic wants to know how much control an agent must have, and when it must tell the user. The report gives no test results.
Sources
Posted
