This legislation amends title 10, United States Code, to establish a comprehensive policy for the Department of Defense (DoD) regarding the acquisition and use of **autonomous weapon systems** and **artificial intelligence capabilities**. The core policy mandates maximizing the use of these technologies while ensuring continuous and rigorous **human oversight** to comply with the law of war, applicable treaties, weapon system safety rules, rules of engagement, and frameworks protecting privacy and civil liberties. The Secretary of Defense is required to ensure personnel exercise appropriate human judgment, consistently monitor deployed systems, and maintain responsibility for their development and use. Other requirements include taking deliberate steps to ensure accuracy, promoting understanding of the technology, and subjecting prototypes and deployed systems to routine **testing and information assurance** across their entire life cycles. Systems must be designed to allow human operators to detect and avoid unintended consequences and retain the ability to intervene or deactivate systems demonstrating unintended or illegal behavior. The bill establishes a detailed **review and verification** process, categorizing systems by their risk to human life, safety, or data integrity. Before prototyping or formal development, the Under Secretary of Defense for Research and Engineering must verify system design incorporates human judgment, engagement parameters, non-target risk accounting, system safety, cybersecurity, and preliminary legal review. A similar, more comprehensive review is required before fielding, covering human-machine interfaces, doctrine, cyber survivability, and an updated legal review. These verifications are valid for three years, and substantially similar variants are also considered verified. The Secretary must ensure subsequent reviews for previously verified or exempted systems without current valid verification. Additionally, **Privacy Impact Assessments** are mandated for high-risk artificial intelligence capabilities, and the requirement for an updated legal review can be temporarily waived in response to real-world conflict or compelling national interest. The legislation outlines specific **exceptions** to these policies and requirements, including operator-supervised systems for time-critical attacks, cyberspace capabilities, unarmed platforms, and unguided munitions. Crucially, it also includes explicit **prohibitions** on certain uses of autonomy or artificial intelligence capabilities. These include decisions to initiate nuclear weapon launches, monitoring or targeting U.S. persons without a warrant (with legal exceptions), and employing lethal force by autonomous weapon systems without appropriate levels of human judgment. The Director for Operational Test and Evaluation is tasked with ensuring rigorous hardware and software **verification, validation, testing, and evaluation** for all covered systems. This includes analysis of unanticipated emergent behavior, iterative cyber testing, and confirmation that autonomy algorithms can be rapidly reprogrammed to correct unintended behaviors. The testing must also ensure adequate training, clear human-machine interfaces, and ongoing monitoring for changes in operational environment or system design. Finally, the bill establishes an "Autonomy and Artificial Intelligence Working Group" to support senior DoD leadership in their review processes, advise on weapon system approvals, and develop safety standards. The Secretary is also required to submit an annual report to congressional defense committees on the administration of this section until January 31, 2037.