S.Amdt. 5968

The amendment, known as the Human Authority in Lethal Operations Act of 2026 (or HALO Act of 2026), aims to add a new subtitle to the bill regarding the use of artificial intelligence in weapon systems. Its primary purpose is to ensure that humans maintain ultimate discretion, judgment, and control over the use of force when the Department of Defense uses autonomous or semi-autonomous weapon systems.

The amendment seeks to change and add the following to the bill:

• Design and Safety Standards: It requires that any system using artificial intelligence to recommend or engage targets must have a clear chain of command. A designated commander must be held accountable for every engagement and must exercise ultimate control over the use of force.

• Strict Review Processes: It adds requirements for pre-development and pre-fielding reviews to verify that systems are safe, legal, and allow for human override.

• Prohibited Uses of AI: It forbids the use of AI for activities that violate the Constitution or civil liberties, such as:

  • Profiling or tracking individuals based on protected rights (like freedom of speech).
  • Inferring an individual's race, religion, gender identity, or political identity.
  • Tracking the location of people in the U.S. without a warrant.
  • Removing humans from the decision-making process for nuclear weapons employment.

• Accountability and Oversight: It establishes reporting requirements to Congress, including semiannual reports on AI use cases and compliance for fielded systems. It also updates whistleblower protections to allow personnel to report concerns about AI misuse.

• Testing and Monitoring: It mandates rigorous testing, including "red-team assessments" to find vulnerabilities, and continuous monitoring to prevent unintended engagements.

• Exceptions: Certain systems are exempt from these specific requirements, including unarmed platforms, unguided munitions, mines, and specific time-critical defensive systems used against incoming rockets or missiles.