S. 4682 — Ultimate Human Responsibility in Defense Systems Act of 2026

This section defines autonomous weapon system as a weapon that, once activated, can select and engage targets without further human intervention. It also defines ultimate human responsibility as the capacity of a human commander or operator to understand a system's operational context, supervise or terminate its use of force, and ensure compliance with U.S. and international law, and defines use of force as the application of kinetic lethal or non-lethal military force against a target.

This section directs the Secretary of Defense to ensure that any autonomous weapon system or artificial intelligence capability developed, acquired, or operated by the Department of Defense enables commanders and operators to exercise ultimate human responsibility over the use of force. It requires design features allowing human supervision, intervention, and fail-safe manual control during mission execution; sufficient information and records for oversight and post-engagement review; and procedures for operator training, proficiency assessments, and anonymous reporting of system reliability concerns.

This section requires the Director of Operational Test and Evaluation to review and certify any autonomous weapon system capable of independently selecting and engaging targets without human supervision before the Department of Defense may field it. The review must include independent adversarial red-team testing under operationally realistic conditions, including electronic warfare and communication disruption, and the certification must confirm ultimate human responsibility for the use of force and secure, safe, and reliable operations with retesting after significant software updates. It also requires the Director to publish unclassified testing standards for artificial intelligence components, including benchmarks for reliability, robustness, security, and human-machine team performance, on a public website within 18 months.

This section requires the Secretary of Defense to submit a report to the congressional defense committees not later than one year after enactment and semiannually thereafter through six years after enactment. Each report must summarize autonomous weapon systems in development or fielded by the Department of Defense, describe steps taken to ensure compliance with the Act, and include any recommendations on maintaining human judgment in such systems.

This section establishes a centralized Department of Defense repository for reporting, collecting, and analyzing incidents involving autonomous weapon systems and artificial intelligence capabilities. It requires reporting of operational malfunctions, unintended behaviors, target selections inconsistent with commander intent or rules of engagement, and human-machine interaction failures; directs modeling of reporting practices on aviation safety systems; mandates analysis and dissemination of trends and lessons learned across the Department; and requires the Secretary to provide semiannual classified briefings on identified risks and corrective actions to the congressional defense committees.

This section directs the Secretary of Defense to develop formal training pipelines for autonomous systems of the Department of Defense. The pipelines must include specialized training tracks for operators, commanders, and system supervisors; certification in human-machine teaming; mandatory training in denied and contested environments such as loss of GPS, electronic warfare, and degraded communications; integration of autonomous systems into Combat Training Center rotations; and continuous retraining tied to software updates.

This section establishes within the Office of the Secretary of Defense the Civilian Harm Mitigation and Response Office to develop, oversee, and coordinate Department-wide policies, practices, and capabilities to prevent, mitigate, and respond to civilian harm resulting from military operations. The Office integrates such practices into planning, targeting, and operations; coordinates assessments and responses to incidents; supports training and doctrine; reports regularly to Congress; and is headed by a senior official designated by the Secretary of Defense.

This section clarifies that the Act does not prohibit the development or use of autonomous weapon systems designed to intercept incoming threats when human supervision remains part of the operational design.