States Fill Regulatory Gap as Federal Government Falters on AI Oversight
States Take the Lead in AI Regulation as Federal Government Steers Clear
As the debate surrounding the regulation of artificial intelligence (AI) technologies reaches a fever pitch at the federal level, state legislatures have stepped forward to fill the gap. In a resounding 2025 trend, all 50 US states have introduced various AI-related legislation aimed at mitigating the challenges posed by these cutting-edge technologies.
While Congress recently voted down a proposed moratorium on state-level AI regulation, states are taking proactive measures to oversee and regulate AI development and deployment within their jurisdictions. This surge in state-led regulation is particularly concerning for AI developers, who must now navigate a complex web of 50 different regulatory frameworks to operate across the country.
From government use of AI to health care, facial recognition, and generative AI, several aspects of AI stand out as critical areas of focus from a regulatory perspective. As we delve deeper into each of these sectors, one thing becomes clear: the lack of federal regulation has created an environment in which states are compelled to take bold action.
Government Use of AI
Predictive AI – that is, AI systems designed to make statistical forecasts and drive predictive models – has significantly transformed many government functions. For instance, AI-driven predictive analytics can help determine eligibility for social services or inform decisions on criminal justice sentencing and parole.
However, the widespread use of algorithmic decision-making by public sector organizations raises serious concerns about bias and accountability. A 2025 review of study findings indicates that racial and gender biases can be programmed into these systems, leading to potentially severe consequences for those affected.
The critical importance of transparency and responsible AI governance is evident in proposals being advanced at the state level. Several states have introduced bills aimed specifically at public sector use of AI, emphasizing the need for accurate disclosure of potential algorithmic harms.
For instance, Colorado’s Artificial Intelligence Act includes provisions requiring developers to disclose risks posed by their systems – both prior to deployment and subsequently through periodic reporting and audits. This regulatory push may help prevent unintended consequences of government reliance on AI-driven predictive models.
In a complementary move, some states have established dedicated bodies to oversee the regulation and development of public sector AI initiatives. New York’s SB 8755 bill is one notable example, establishing an expert review board responsible for ensuring that AI development aligns with regulatory guidelines.
AI in Healthcare
The intersection of healthcare and AI has become increasingly significant, driving the introduction of more than 250 proposed bills aimed at regulating AI in this sector within a matter of months. The majority of these healthcare-centered bills fall into one or more of four categories: disclosure requirements, consumer protection measures, insurers’ use of AI, and clinician reliance on AI-driven diagnostic tools.
Bills promoting transparency require developers to disclose critical information regarding system functionality and any potential risks to patient safety. Consumer protection provisions aim to prevent AI-driven systems from unfairly discriminating against certain populations, and ensure that patients have meaningful avenues for appealing decisions made through the use of AI.
Legislation focused on insurers’ usage addresses the oversight needed in areas such as insurance approval and payment decision-making processes, where AI’s potential for bias may manifest. Meanwhile, proposals regulating clinician reliance on AI-driven technology to diagnose and treat patients aim to ensure the responsible application of these groundbreaking tools within medical settings.
Facial Recognition and Surveillance
One area of particular concern – and subject to extensive state-level legislation – is facial recognition software. Historically viewed as a tool useful in both predictive policing and national security sectors, recent studies have highlighted significant privacy risks associated with its deployment, particularly in areas such as bias against darker-skinned individuals.
Concerns over facial biometrics technology led 15 US states by the end of 2024 to introduce legislation aimed at mitigating these challenges. Various elements of state-level regulations attempt to address issues like bias through:
* Vendor transparency requirements for publishing test reports on bias
* The need for human oversight in AI-driven surveillance and facial recognition applications
