Where needed, formulate and publish policies such as the actor's AI governance policy and privacy policy for risk management and safety of AI use (including the social responsibility of sharing a vision with society and the public and communicating information); where needed, create opportunities to receive stakeholders' reports of errors in AI output and carry out periodic, objective monitoring; and formulate a policy for how to respond when stakeholders' interests are harmed, implement it reliably and report progress to stakeholders regularly as needed. The guidelines add that inquiries or disclosure requests from rights holders of training data or people disadvantaged by AI judgments should be handled appropriately within the necessary range, and court disclosure orders followed.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.