Public agencies increasingly rely on AI to deliver public services like education, housing, public benefits, and healthcare. As state governments expand their investment in this technology, it is critical that they do so in ways that encourage responsible use. The following three policy priorities, along with emerging state legislative examples, are important steps toward establishing meaningful, robust practices around trustworthy and safe AI use.
Three Policy Priorities for State Legislation
Legislation should require that, when adopting AI, public agencies mitigate risks, stand up durable decision-making structures, and increase access to vital information.
Risk Management: Implementing the following robust guardrails on high-impact uses that affect individuals’ rights or safety ensures that AI tools are well-vetted before and during their deployment, which protects the public and government agencies from potential harms:
- Establish heightened requirements for high-impact systems, such as pre- and post-deployment assessments, human oversight, emergency protocols, and independent accountability.
- Incorporate AI governance requirements in contracts with government service providers, such as contract provisions requiring risk management practices and data privacy protections.
- Require public consultation throughout the AI lifecycle, including before agencies deploy high-impact tools, and include representation from civil society, academia, and impacted communities in state-wide task forces.
- Direct agencies and government-wide personnel offices to develop and administer employee training on the responsible use of AI tools, including privacy and cybersecurity protections.
AI Governance: Creating the following durable, centralized leadership structures for overseeing AI use and deployments helps agencies make the best use of their resources and coordinate AI projects across agencies:
- Establish centralized oversight and leadership structures by designating an office and/or chief AI officer (or another equivalent senior leader) responsible for managing the state’s use of AI, including creating acceptable use policies for AI tools and promulgating rules necessary to implement statutory provisions.
- Create a government-wide AI governance board to guide statewide AI governance priorities.
Transparency: Providing the public with the following clear, accessible information about the AI tools used by government agencies builds public trust and facilitates effective oversight:
- Establish public AI inventories that document and disclose how AI is being used, tested, and governed across all state agencies, and require that they are updated on a regular basis.
- Institute notice and disclosure requirements for public-facing AI tools, including explanations about why and how an outcome was determined or influenced by an AI system.
Emerging State Legislative Examples
States across the political spectrum are enacting new guardrails that enable safe, trustworthy AI use by public agencies:
- Maryland’s SB 818 imposes strong guardrails on state agencies’ uses of AI, requiring agencies to conduct impact assessments and publicly report about any high-impact AI systems. The bill also establishes the Governor’s Artificial Intelligence Subcabinet, which is responsible for developing policies and procedures for state agencies to conduct ongoing monitoring of AI systems.
- Kentucky’s SB 4 establishes a comprehensive approach to public sector AI governance. Its requirements include directing the Office of Technology to establish standards for the responsible use of AI (including risk management policies for high-impact AI systems), establishing an AI Governance Committee, requiring agencies to publicly disclose their use of AI, and creating an AI inventory.
- Texas’ SB 1964 creates a strong government-wide framework for public sector AI use and oversight. The bill’s provisions require the Department of Information Resources to inventory AI systems deployed by state agencies, require state agencies to adopt risk management and governance standards for high-impact AI systems, and direct the Department to establish an AI code of ethics that addresses human oversight, accuracy, privacy, and security.
- Montana’s HB 178 establishes some key safeguards around how public agencies use AI. This includes requirements for government agencies to disclose the use of AI, human review obligations for high-impact systems, and a prohibition on the use of AI by government agencies for cognitive manipulation, social classification, deception, and surveillance in public spaces.
For more info on CDT’s efforts to advance state legislation that governs AI use by public agencies, visit https://cdt.org/area-of-focus/equity-in-civic-tech/ai-in-public-benefits/ or email Travis Hall at [email protected].
Facts Only
Public agencies are using AI to deliver services in education, housing, public benefits, and healthcare.
State governments are expanding investments in AI technology.
Three policy priorities for responsible AI use are risk management, AI governance, and transparency.
Risk management includes pre- and post-deployment assessments, human oversight, and public consultation for high-impact AI systems.
AI governance involves centralized oversight structures, such as a chief AI officer or a statewide governance board.
Transparency measures require public AI inventories and disclosure of AI tools used by government agencies.
Maryland’s SB 818 mandates impact assessments and public reporting for high-impact AI systems.
Kentucky’s SB 4 establishes AI governance standards and requires public disclosure of AI use.
Texas’ SB 1964 creates a government-wide framework for AI oversight, including risk management and an AI code of ethics.
Montana’s HB 178 includes human review requirements for high-impact AI systems and prohibits AI use for cognitive manipulation.
The Center for Democracy & Technology (CDT) is involved in advancing state legislation for AI governance in public agencies.
Executive Summary
State governments are increasingly adopting AI to deliver public services, prompting calls for responsible governance to mitigate risks and ensure transparency. Three key policy priorities are emerging: risk management, AI governance, and transparency. Risk management involves heightened requirements for high-impact AI systems, including pre- and post-deployment assessments, human oversight, and public consultation. AI governance emphasizes centralized oversight structures, such as appointing a chief AI officer or establishing a statewide governance board. Transparency measures include public AI inventories and disclosure requirements for AI tools used by government agencies.
Several states have enacted legislation to address these priorities. Maryland’s SB 818 requires impact assessments and public reporting for high-impact AI systems, while Kentucky’s SB 4 mandates AI governance standards and public disclosure. Texas’ SB 1964 establishes a government-wide framework for AI oversight, including risk management and an AI code of ethics. Montana’s HB 178 includes safeguards like human review for high-impact systems and prohibitions on AI use for cognitive manipulation. These legislative efforts reflect a growing recognition of the need for structured oversight to balance innovation with public trust and safety.
Full Take
The narrative presents a compelling case for structured AI governance in the public sector, emphasizing risk mitigation, centralized oversight, and transparency. The strongest version of this argument is that proactive legislation can prevent harm while fostering innovation, a balance that aligns with broader societal concerns about AI’s ethical implications. The examples from Maryland, Kentucky, Texas, and Montana demonstrate bipartisan recognition of the need for guardrails, suggesting a rare consensus on the urgency of the issue.
However, the discussion assumes that legislative frameworks alone can address the complexities of AI deployment. It does not explore potential challenges, such as enforcement gaps, bureaucratic inertia, or the risk of overregulation stifling beneficial innovation. The focus on high-impact systems also raises questions about how "impact" is defined and who gets to decide—could this lead to arbitrary exclusions or over-inclusion? Additionally, while public consultation is advocated, the effectiveness of such processes in influencing policy outcomes remains uncertain.
Root cause: This narrative reflects a broader paradigm shift toward treating AI as a public good requiring democratic accountability. It echoes historical patterns of technological regulation, where initial enthusiasm gives way to caution once societal risks become apparent. The underlying assumption is that government can—and should—act as a steward of ethical AI, but this presumes both capacity and political will, which may vary widely across states.
Implications: If successful, these policies could enhance public trust in government AI use, but they may also create compliance burdens for agencies. The costs of implementation—financial, administrative, and operational—could fall disproportionately on smaller or under-resourced states. Second-order consequences might include a chilling effect on AI adoption in the public sector or the emergence of a patchwork of state-level regulations that complicate interstate cooperation.
Bridge questions: How might these policies interact with existing privacy laws or civil rights protections? What mechanisms could ensure that public consultation processes are meaningful rather than performative? Would a federal framework be more effective than state-by-state regulation, or does decentralization allow for necessary flexibility?
Counterstrike scan: A coordinated influence campaign pushing this narrative might exaggerate risks to justify expansive regulatory powers or frame opposition as technologically illiterate. However, the content here aligns with legitimate governance concerns rather than manipulative tactics. The focus on bipartisan examples and practical policy steps suggests a good-faith effort to address a complex issue.
Patterns detected: none
Sentinel — Human
This analysis indicates that the article is likely to be written by a human. The text shows variance in sentence structure and length, exhibits balanced framing with a clear argumentative structure, but does not exhibit any apparent matching of argumentative skeleton or template patterns. Human signals include an understanding and application of complex ideas.