How U.S. States Are Outpacing the Feds on AI Rules—and Where Enforcement Stalls
The U.S. has no federal AI law, and with the Department of Justice scaling back oversight, states are stepping into the regulatory void. California and New York are leading with tailored laws, but a patchwork of approaches risks leaving gaps in safety and accountability. Meanwhile, the EU’s AI Act offers a model no U.S. state has fully matched.
The shift is clear: while Washington debates, state legislatures are drafting rules for everything from AI in hiring to healthcare. But without dedicated agencies or funding, enforcement remains a weak link. The question now is whether these efforts will harmonize—or fragment into a regulatory maze.
The Patchwork of State AI Laws
California’s AI Bill of Rights, passed in 2023, set principles for transparency, accountability, and bias mitigation. But it lacks enforcement teeth, leaving companies to self-regulate. New York’s AI Task Force, formed in 2024, proposed guidelines for AI in hiring, healthcare, and public services—echoing some categories in the EU’s AI Act, which classifies systems by risk.
Other states are moving in different directions. Texas and Florida have passed laws restricting AI regulations, prioritizing industry flexibility over consumer protections. Meanwhile, Washington and Illinois have proposed AI-specific enforcement mechanisms, though funding and expertise remain barriers.
The numbers tell the story: no U.S. state has adopted a comprehensive AI framework like the EU’s, which imposes strict rules on high-risk applications like healthcare and autonomous vehicles. Instead, most laws focus on narrow sectors—California’s AI in education, New York’s AI in policing—leaving gaps in broader oversight.
How State Laws Compare to the EU’s AI Act
The EU’s AI Act is the gold standard, with risk-based classifications and real-time monitoring for critical systems. No U.S. state has replicated this structure. California’s principles are aspirational, while New York’s guidelines are advisory. Even where states align—such as in healthcare—enforcement is inconsistent.
Take medical AI. The EU Act would require strict oversight for AI used in diagnostics or treatment. In the U.S., no state has adopted real-time monitoring or bidirectional patient-AI interaction systems, despite research showing potential in epilepsy management. A 2026 study in Nature Medicine highlighted how implantable devices could alert patients to seizure risks—but none yet share data in real time.
The trend underneath is a race between innovation and regulation. States are acting, but without federal coordination, the risk is a fragmented system where companies exploit loopholes. The EU’s approach—clear categories, enforceable rules—remains the exception, not the rule.
Enforcement: The Weak Link
States lack dedicated AI regulatory agencies. California and New York rely on existing bodies like consumer protection offices, which lack the expertise to police complex AI systems. Washington and Illinois have proposed stronger enforcement, but funding and staffing remain hurdles.
The DOJ’s retreat from AI oversight in 2026 shifts pressure to states, but coordination is minimal. Texas and Florida’s laws reflect industry pushback, while progressive states struggle to keep up with rapid AI advancements. The result? A regulatory free-for-all where companies pick the jurisdiction with the weakest rules.
Watch this space: if states don’t harmonize, the U.S. could end up with a patchwork where AI safety depends on where you live. The EU’s AI Act shows what’s possible—but without federal leadership, the U.S. risks falling behind.
What Comes Next?
The next phase will test whether states can collaborate or if fragmentation wins. California and New York may lead, but without federal standards, enforcement will stay uneven. The EU’s AI Act proves comprehensive rules work—but the U.S. is still figuring out how to make it happen.
What’s changing? States are drafting laws, but the real test is whether they can enforce them. The trend underneath is clear: the U.S. is moving forward, but without a unified approach, the risks outweigh the rewards.
Real-Time Data Sharing in Medical AI Devices
Current implantable epilepsy devices can store and transmit data asynchronously but do not share it in real time, according to a 2026 study in Nature Medicine. This limits the ability to convey useful information such as changes in seizure risk or detection of seizure precipitants like missed medications or alcohol consumption. The study notes that no existing device offers bidirectional interaction that allows patients to communicate with the device in real time.
New minimally invasive EEG devices are being developed to continuously collect data during daily life and transmit it to smartphones and the cloud for clinical viewing. Wearable versions monitor movement, heart rate, and other peripheral parameters to alert families to convulsive seizures. However, none of these devices currently provide bidirectional interfaces for real-time patient-device interaction.
The research emphasizes that no prior literature describes interactive epilepsy devices with bidirectional communication capabilities. Early work focused only on warning of impending seizures or forecasting seizure-free periods. An optimal device, as proposed in the study, would present sensor-derived EEG data intelligibly and answer patient queries with actionable feedback based on real-time analysis.
NASA’s Role in AI-Driven Space Exploration and Disaster Response
NASA has integrated “super intelligence” (SI) into its missions, particularly in autonomous navigation and data analysis, as outlined in the **Sept. 29, 2026, Executive Order 14434**, which directs federal agencies to adopt the term. The agency’s **Perseverance rover**, for instance, relies on SI for 88% of its driving, using onboard cameras and AI to analyze terrain, identify hazards, and navigate autonomously—even on uncharted Martian landscapes. This capability extends to deep-space missions, where AI enables spacecraft to make critical decisions without Earth communication, such as adjusting trajectories or troubleshooting system failures.
Beyond exploration, NASA leverages SI to enhance disaster response. By applying AI tools to satellite imagery, the agency detected the number of **blue tarps on rooftops** post-hurricane as a proxy for damage severity in affected communities. This real-time data helps prioritize relief efforts, demonstrating how AI can bridge gaps in emergency coordination. Additionally, NASA’s **Digital Information Platform (DIP)** uses SI to optimize commercial flight routes, reducing fuel consumption and improving airspace safety by autonomously generating efficient, collision-avoidance paths for aircraft.
The agency’s collaboration with private-sector AI leaders—including partnerships in the **space economy and commercial space sector**—further accelerates innovation. For example, SI aids in analyzing vast datasets from telescopes to identify exoplanets or model climate trends, while autonomous systems support **Moon and Mars missions** by handling routine tasks, freeing human operators for higher-level oversight. However, NASA’s use of AI remains mission-specific, with no unified framework for broader regulatory alignment across federal or state jurisdictions.
Sources
- Artificial intelligence
- Super Intelligence – NASA
- Human–artificial intelligence collaboration for comanaging epilepsy in real time – Nature Medicine
Photo: Daniel Schwen (CC BY-SA 4.0)
