Can Agents Be Safe in the Open? Two Levels of Trust
Audrey Tang and Vitalik Buterin climb toward that question one level at a time, taking the questions the room upvotes in the Meerkat they share with Audrey's talk, which runs just before.
Level one covers everyday failures: an agent that makes a mistake, or one tricked from outside. Level two is the hard case, where no one can simply take the model makers' word, and asks what anyone can still check.
Speakers: Audrey Tang, Vitalik
Track: Rights, Freedoms, and Governance
