There are now at least five active efforts to build trust infrastructure for AI agents, and none of them are interoperable. That's not a coordination failure. It's a signal about what "trust" actually means.