There is no central authority. The Trust Score is a conceptual framework, not a shared database. Each AI platform (OpenAI, Anthropic, Google) builds its own model. They retain full discretion. Agents do not talk to each other. They talk to users. If a score is low, the agent warns the user. It adds caveats or drops the recommendation. It does not broadcast to other bots.