Do not develop, provide or use AI systems or services whose purpose is to unduly manipulate people's decisions, cognition or emotions, or which presuppose manipulation below the level of conscious perception. Pay attention to the risk of over-reliance on AI such as automation bias and take the necessary countermeasures (the footnoted examples: training so that outputs are not accepted blindly, and people forming their own reasons before approving an AI judgment). Pay attention to uses that promote a skew of information or values, such as a filter bubble, and unwillingly narrow the options people should have. Handle AI outputs with particular care where they may bear on procedures with a serious social effect, such as elections and community decision-making.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.