AI Agent Economy Needs Dispute Resolution Infrastructure

Federico Ast's Ethereum Foundation talk examines governance for an economy in which AI agents transact directly with each other, including negotiating, paying and contracting. The description says this economy is arriving fast, but governance infrastructure is not yet in place. Many disputes will occur at machine speed and involve amounts too small for a human court case. The talk maps out two needs: certifying which agents can be trusted to transact, and resolving disputes through agent juries that rule on agent-to-agent cases. It also reports that in real experiments different frontier models ruled differently on the same disputes: Claude was nearly 5x more likely than ChatGPT to rule in the consumer's favor. Early signs suggest agents can carry human-like biases into decisions. The talk asks what neutral dispute resolution for agents looks like and what mechanisms could achieve it.