This legislation establishes a comprehensive framework for the Department of Defense's (DoD) use of artificial intelligence, aiming to ensure its secure, accountable, and responsible deployment. It specifically addresses high-consequence artificial intelligence applications , which are defined as those presenting heightened national security, operational, safety, legal, or civil liberties risks. The bill mandates a process for designating such applications, including AI for nuclear weapons, lethal targeting, cyber operations, and autonomous weapon systems. For high-consequence AI applications, the bill requires approval by a senior DoD official before operational deployment. This approval is contingent upon satisfying minimum requirements such as realistic testing and evaluation , legal and policy review, clear documentation of intended use, and established fallback procedures. Furthermore, the legislation mandates human accountability , ensuring that AI supports, but does not substitute for, human judgment in critical decisions like the use of force. Congress must also be notified prior to initial operational deployment, with an annual briefing on implementation. The bill imposes strict reporting requirements on contractors providing frontier AI models to the Department. Covered contractors must report various covered incidents , including theft of model weights, foreign adversary attempts to compromise AI systems, supply chain compromises, and concerning model behaviors like increased cyber offense capabilities or deceptive actions. Initial reports for critical incidents are due within 72 hours, with others due within 7 days, and the Department must then notify Congress within 7 days of receipt. Significant limitations are placed on certain AI uses by the Department. AI is prohibited for selecting targets for nuclear weapons or their detonation, and generally for domestic person-based analysis of U.S. persons, with specific exceptions for cybersecurity or force protection. The bill also restricts the development or employment of autonomous weapon systems , allowing only semi-autonomous systems or operator-supervised autonomous systems for local defense under strict conditions. These excepted systems must undergo legal review, rigorous testing, and ensure appropriate human judgment and control. For any autonomous weapon system not meeting these exceptions, the Secretary of Defense can petition Congress for authorization. Such a request must include a detailed military necessity statement, operational constraints, a legal opinion on compliance with the Law of Armed Conflict, and a risk mitigation plan. Congressional approval requires a joint resolution, and any authorization granted would expire after three years, necessitating renewal.