Bill > S4707

This bill, titled the "Responsible Artificial Intelligence Defense Act of 2026," establishes a policy for the Department of Defense (DoD) to maximize the use of autonomy and artificial intelligence (AI) systems while ensuring continuous and rigorous human oversight to comply with the law of war, treaties, safety rules, rules of engagement, and privacy protections for U.S. persons. It mandates that the Secretary of Defense ensure personnel exercise appropriate human judgment, monitor deployed AI-enabled autonomous weapon systems, and remain responsible for their development and use. The bill requires deliberate steps to ensure accuracy in these systems, promotes understanding of their technology and operations, and mandates routine testing and information assurance throughout their lifecycles. It also requires that these systems be designed so human operators can detect and avoid unintended consequences and intervene to disengage systems exhibiting unintended or illegal behavior. Furthermore, the bill outlines a review and verification process for autonomous weapon systems and supporting AI capabilities, categorizing them by the level of human judgment required to mitigate risks, and mandates reviews before prototyping, formal development, and fielding, including legal reviews and privacy impact assessments for certain AI capabilities. It also specifies exceptions for certain types of systems, such as unguided munitions or unarmed platforms, and prohibits the use of AI for initiating nuclear weapon launches or for certain surveillance activities within the U.S. without a warrant, and for lethal force by autonomous weapon systems without appropriate human judgment. The bill also establishes an "Autonomy and Artificial Intelligence Working Group" to advise on these matters and requires an annual report to Congress.