Peters and Lankford introduce bill requiring AI notices and human review at federal agencies

Federal agencies would have to tell people about certain AI interactions and offer human review of adverse critical decisions under legislation introduced October 5, 2026. Senators Gary Peters and James Lankford’s bipartisan Transparent Automated Governance (TAG) Act pairs disclosure with a route to challenge decisions, rather than stopping at notice that automation was involved.

The proposal responds to technology already being used in public services. The Senate Homeland Security and Governmental Affairs Committee’s announcement says federal agencies use AI both to interact with the public and to make decisions affecting individuals. It also expects that use to grow.

Peters, a Michigan Democrat and the committee’s ranking member, framed transparency as an obligation to people relying on government services. The announcement argues that people who unknowingly encounter AI can be confused or frustrated by its determinations. Lankford, an Oklahoma Republican, emphasized security, privacy and unintended consequences alongside the technology’s potential to improve government efficiency.

OMB would guide two kinds of notice

The bill would direct the director of the Office of Management and Budget, or OMB, to issue guidance on transparency practices for AI and other automated systems. That makes OMB guidance the mechanism for directing agencies to implement the proposed requirements. The framework covers both an encounter with an automated system and the use of AI in a critical decision about someone.

• Interaction notice: Agencies would notify individuals when they are interacting with certain automated systems. This part of the proposal addresses the public-facing encounter itself, rather than depending on whether that interaction produces a critical decision about the person.
• Decision notice: Agencies would also notify individuals when a critical decision about them uses an AI-augmented decision process—one in which AI helps make the decision. The proposal therefore addresses AI-assisted decisions, not only decisions described as being made entirely by an automated system.

Artificial intelligence is a powerful tool to improve the efficiency and effectiveness of our federal government, but we must always keep security, privacy, and any unintended consequences in mind before we turn any process or decision over to a computer

Sen. James Lankford, in the committee announcement

A way to contest the outcome

The proposed response to an adverse critical decision goes beyond disclosure. OMB’s guidance would instruct agencies to establish redress processes—ways for affected people to challenge an outcome—that include human review. That requirement is tied to critical decisions that negatively affect individuals, rather than a promise of human review for every routine interaction with an automated federal service.

The requirements remain proposed, not rules taking effect with the bill’s introduction. Their reach also matters: the sponsors describe notice for “certain” automated systems and review for “critical” decisions. The committee announcement does not define those boundaries, leaving the exact set of covered interactions and decisions unclear in the sponsors’ public summary.