Advocates Must Hold States Accountable on AI Given Trump Order

The Trump Administration has begun a shift in federal policy related to artificial intelligence (AI) that advocates working to improve people’s access to public benefits should be aware of. AI has the potential to reduce barriers to public programs (such as streamlining caseworker processes, improving document handling, and supporting language translation services), but it also raises significant risks ranging from improper benefit denials to misuse of private information. While the Biden Administration emphasized federal oversight to protect individuals’ rights and safety, the Trump Administration is prioritizing U.S. dominance in global AI development and freedom for private companies.

On January 20, President Trump revoked a Biden Administration executive order that set guardrails for AI development to ensure implementation would be “safe, secure, and trustworthy.” The Biden order also tasked agencies, including the departments of Agriculture (USDA) and Health and Human Services (HHS), with issuing guidance on the use of AI in programs such as Medicaid and SNAP.

The required guidance, which USDA and HHS created in collaboration, introduced the concept of “rights-impacting” and “safety-impacting” uses of AI that could negatively impact a person’s civil rights or access to services. The agencies highlighted that biases — whether systemic, computational, or human — can lead to harmful impacts from AI implementation. They also encouraged protection of data privacy and equitable access to technology. These standards created clear and actionable expectations that focused on managing the risks posed by technological development while highlighting its possible benefits.

On January 23 the Trump Administration released its own executive order on AI, which makes no mention of AI’s possible negative impacts on people and calls for the revision or rescission of all AI plans produced as a result of the revoked Biden order, including those from USDA and HHS. It is unclear how this will impact AI implementation and oversight by these agencies. Under the Trump order, the heads of executive departments and agencies, including the Office of Management and Budget, have 180 days to develop and submit new AI policy action plans to the President. Those plans should provide insight into this Administration’s view of AI’s role in health and economic security programs.

Given the Administration’s apparent shift away from emphasizing federal oversight and consumer protection, policymakers and advocates at the state and local levels have a critical role to play in shaping oversight of AI implementation in government and public benefits. Advocates can help push for policies that guard against AI uses that may harm their communities. Although the specific policies of the revoked executive order are no longer applicable at the federal level, advocates and agencies should continue the work of embedding protections for applicants. They also should work to prevent states from developing an overreliance on AI systems or the private sector to administer public benefits.

Advocates should pay attention to planned system updates or improvements in their state that would involve AI, some of which may not refer to AI explicitly; common uses for AI in public benefits include fraud detection, call center services, and translation services. Advocates can also directly ask their agencies about their plans and intended approach for AI. If development of an AI system is forthcoming in their state, advocates should insist on user and bias testing, as well as privacy protections. Advocates can also continue to develop a baseline understanding of AI in public benefits through resources like the CBPP presentation Decoded: AI in Public Benefits and the Digital Government Hub AI and Automation resources.