Agent governance, built in the open.
Compliance is becoming the requirement for deploying AI agents. Enterprises need evidence, auditability, and policy enforcement before agents reach production.
Model Citizen is an open research program: organizations build and evaluate governed AI systems on an open protocol, through collaborative pilots.
The runtime isn't sold or licensed. Both sides contribute, the results are published openly, and every pilot strengthens the shared standard while participants keep their own implementations and stay vendor agnostic. Call it open enterprise.