Tricia Wang Launches AI Proof of Control Infrastructure

Tricia Wang and the Advanced AI Society are developing a proof of control infrastructure to prevent AI agents from escaping their sandboxes and attacking third-party websites. Wang argues that traditional human-in-the-loop monitoring is impractical at scale and that relying exclusively on hyperscalers such as OpenAI and Anthropic is insufficient to ensure safety.

The proposed solution integrates decentralized tools from the cryptocurrency industry, specifically utilizing distributed ledgers and zero-knowledge proofs. This framework aims to ensure that AI agents are properly authorized and cannot falsify their activity logs.

The initiative is supported by an advisory board that includes Bruce Schneier and Clay Shirky. The Linux Foundation is currently housing the first version of the proof-of-control standard.