S. 4697 — HALO Act of 2026

This section defines terms used in the Act, including autonomous weapon system, semi-autonomous weapon system, covered artificial intelligence capability, unintended engagement, designated commander, and specific target group.

This section establishes design, testing, development, procurement, deployment, and use requirements for autonomous and semi-autonomous weapon systems that employ artificial intelligence to identify, recommend, or engage targets in support of use-of-force decisions. It requires a designated commander to exercise ultimate discretion and accountability over each engagement or class of engagements; mandates system features that limit actions to designated timeframes, geographic areas, and target sets, require human review before expanding those parameters, and terminate engagements when parameters cannot be met; imposes record-keeping, auditability, explainability, cybersecurity, and human-machine interface standards; and directs public availability of the Department’s Ethical Principles for Artificial Intelligence and Responsible Artificial Intelligence Strategy and Implementation Pathway, with 30 days’ advance notice of revisions.

This section requires the Secretary of Defense to obtain joint verification from designated senior officials before formal development and before fielding of specified weapon systems. The pre-development review must confirm human commander discretion over use of force, operation within designated parameters and timeframes, risk accounting under the Law of Armed Conflict, safety and cybersecurity measures, testing plans, consistency with AI ethical principles, and completed legal analysis. The pre-fielding review must confirm demonstrated human-machine interfaces and training, implemented safety and monitoring protocols, validation under realistic conditions, and updated legal compliance. Modified variants require re-review if algorithms, missions, operational environments, or targets change.

This section directs the Chief Digital and Artificial Intelligence Officer to monitor and evaluate artificial intelligence capabilities and cybersecurity in autonomous and semi-autonomous weapon systems and advise the Secretary of Defense on such matters. It requires the Officer, in collaboration with the Under Secretary of Defense for Research and Engineering, to formulate testable requirements and issue policy and guidance to implement the Ethical Principles for Artificial Intelligence and the Responsible Artificial Intelligence Strategy and Implementation Pathway, including red-team assessments of weapon system vulnerabilities. It also requires coordination with the Under Secretary and the Director of Operational Test and Evaluation to develop common tools, infrastructure, and assurance benchmarks for test and evaluation.

This section requires the Secretary of Defense to ensure that specified weapon systems undergo rigorous hardware and software verification and validation, realistic developmental and operational testing against adaptive adversaries with civilian presence, and analysis of emergent behavior. It mandates quarterly cyber testing, post-fielding data collection and retesting after any changes, and AI-specific validation of robustness and rapid reprogrammability of autonomy algorithms. It directs the Director of Operational Test and Evaluation to set standards, review test plans, maintain a centralized incident repository, and coordinate monitoring to prevent unintended engagements.

This section prohibits the Department of Defense from using any covered artificial intelligence capability in violation of the Constitution, Federal law, the Law of Armed Conflict, or U.S. treaty obligations, or in any manner posing an unacceptable risk to individual safety or civil liberties. It further bars such use to profile individuals based on protected constitutional rights; infer emotional states; determine religion, race, sexual orientation, immigration status, disability, gender identity, or political identity; track U.S. persons’ locations without a judicial order or warrant; aggregate data to achieve any of these ends; remove humans from nuclear employment decisions; or obtain personal data of U.S. persons from data brokers or other entities if acquired unlawfully. The bans apply to all Department activities and extend to joint operations and any recipient agency when data or systems are shared.

This section directs the Secretary to update whistleblower protections for personnel who develop, assess, deploy, operate, or use artificial intelligence in National Security Systems or for military or intelligence purposes. It requires procedures allowing reports of concerns about harms to civil liberties, privacy, safety, or statutory compliance, including anonymous and confidential channels, investigations of AI misuse, and corrective actions, while complying with existing military, civil service, and intelligence whistleblower statutes.

This section requires the Secretary to submit semiannual reports describing exemplary artificial intelligence use cases in the Department, after-action assessments of significant operational uses that include civilian casualties and Law of Armed Conflict incidents, and operator training; an annual report identifying barriers to responsible AI deployment and infrastructure gaps for traceability and forensics; and semiannual reports on each non-compliant fielded system with operational justifications and remediation timelines.

This section exempts autonomous or semi-autonomous cyberspace capabilities, unarmed platforms, and non-weapon autonomous systems from the requirements of sections 3, 4, 5, and 7. It also exempts time-critical defensive systems against incoming munitions, rockets, artillery, mortars, or missiles, and unmanned aircraft systems, when human authorization is not feasible and no death or serious bodily harm is a reasonably foreseeable consequence. The section further exempts unguided munitions, manually guided munitions, mines, unexploded ordnance, and systems whose sole purpose is identifying potential targets without further human intervention.

This section delays the effective date of the Act until 180 days after enactment. It requires the Secretary to review all Department systems described in section 3(a)(2) that are then in use, submit an attestation to the appropriate congressional committees on whether each complies with the requirements of sections 3, 6, and 7, and for any noncompliant system immediately cease its use while submitting a remediation plan that identifies the unmet requirements, any operational or national security justification for continued use, and a timeline for compliance or retirement.