Washington — A House committee today advanced the Human Decision Standard, legislation backed by the Stewards that would identify categories of federal authority that may not be delegated to autonomous systems, AGI or Cognitive Cores.
Agencies may use advanced systems to model outcomes, identify patterns, prepare options and carry out authorized actions. The bill draws a statutory line between judgment and execution. Where federal action could take a life, deny liberty, initiate military force, impose a binding legal disability or materially redirect national economic policy, a named human official must make and authenticate the decision.
The legislation also requires agencies to maintain a decision record identifying the human authority responsible for each covered action. Systems may generate supporting analysis, but the record may not list an artificial agent as the final decision-maker.
Stewards describe the proposal as an implementation of the Human Imperative: humans occupy the judgment layer unilaterally, while technology operates in the execution layer. The standard preserves an identified human decision-maker while allowing agencies to use automation for speed, accuracy and capacity.
The measure now moves to the full House. Separate language under discussion in the Senate would apply similar requirements to defense acquisition and strategic command systems.
The covered categories focus on consequential exercises of sovereign power. Ordinary clerical processing, logistics, scheduling and authorized automated execution remain available to agencies.
info@stewardsofamerica.com