States are racing ahead of Congress to regulate artificial intelligence, with over 1,500 AI bills currently under consideration in statehouses across the country. This surge represents a dramatic acceleration in AI governance, building on 150 AI-related bills that passed into law in 2025 alone. While Congress debates a bipartisan federal approach, the Trump administration is simultaneously pursuing a deregulatory agenda that seeks to curb state AI lawmaking through executive order.

What Is Congress Proposing to Fix the AI Regulation Mess?

On June 4, 2026, Representatives Jay Obernolte (R-California) and Lori Trahan (D-Massachusetts) introduced the Great American AI Act of 2026, which aims to create a unified federal framework for frontier AI governance. The bill would nationalize the governance approach emerging in several states, though it would preempt certain state laws regulating AI development for only a three-year period. Notably, the current discussion draft would leave much of the existing state patchwork intact, suggesting that federal action may not fully resolve the fragmentation problem.

The congressional approach differs significantly from Europe's model. The European Union's AI Act takes a comprehensive approach that prohibits certain uses of AI deemed unacceptable and includes varying frameworks for "high-risk" and "general purpose" AI systems. By contrast, U.S. regulation has focused on specific issues and sectors rather than a blanket prohibition strategy.

How Are States Actually Regulating AI Right Now?

The emerging consensus on U.S. AI regulation centers on several key themes that states have begun to implement independently:

• Transparency Requirements: States require companies to disclose when consumers are interacting with AI rather than humans. Utah's AI Policy Act, for example, mandates that regulated occupations like healthcare professionals disclose generative AI use at the beginning of interactions, while other businesses must "clearly and conspicuously" inform consumers when asked.
• Companion AI Safeguards: California, Connecticut, New York, Oregon, and Washington have begun regulating "companion AI" services that simulate ongoing human-like relationships. These laws require companies to inform users they are engaging with AI, implement protections against self-harm and suicidal ideation, and include stricter safeguards for minors, such as Oregon's requirement that chatbots remind young users to take breaks every three hours.
• Restrictions on AI in Mental Health: Illinois's HB 1806 sets a critical benchmark by prohibiting AI from replacing licensed therapists in delivering care, while allowing AI to support administrative tasks like scheduling without patient consent. Licensed providers must obtain patient consent for supplementary uses like recording or transcribing sessions.
• Consequential Decision Protections: States including Colorado, Texas, California, Illinois, and New York City have advanced laws requiring transparency, opt-out rights, human review opportunities, and risk management policies when AI is used in employment, financial services, healthcare, education, and housing decisions.
• Training Data Transparency: Some laws, such as California's AB 2013, require developers of generative AI systems to publicly post documentation about their training data and any substantial modifications to those systems.

Colorado's 2026 legislative pivot illustrates how this regulation is evolving. The state replaced its landmark Colorado AI Act with the Colorado Automated Decision-Making Technology (ADMT) Act, broadening the focus from "artificial intelligence" to "automated decision-making technology" more broadly. The new law expands coverage to a wider range of digital technologies and replaces many governance requirements with a framework centered on transparency, consumer rights, and accountability, while introducing a new liability regime designed to encourage companies to implement internal safeguards.

What Is the Trump Administration Doing About State AI Laws?

President Trump's executive order aimed to block state AI lawmaking in order to promote AI innovation and U.S. competitiveness. However, a more recent executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security" has adopted a voluntary standard for frontier AI safety that mirrors the state model and congressional bill approach. This suggests that even the administration's deregulatory push is not entirely displacing the framework emerging at the state level.

The tension between federal preemption and state innovation reflects a fundamental challenge in AI governance. While a unified federal standard could reduce compliance burden for companies operating across multiple states, the current patchwork approach allows states to experiment with different regulatory models and respond to local concerns about AI safety and ethics.

As lawmakers continue to debate the best path forward, the sheer volume of state bills under consideration suggests that AI regulation will remain fragmented for the foreseeable future, even if Congress passes a federal framework. Companies developing and deploying AI systems will need to navigate an increasingly complex landscape of overlapping state and federal requirements, making comprehensive governance strategies essential for compliance and competitive advantage.