States vs Washington: The Growing Battle Over AI Rules in Health Care

Introduction: A Quiet Regulatory War Around Medical AI

Artificial intelligence is moving fast inside hospitals, insurance systems, and mental health platforms, often faster than lawmakers can keep up. While former President Donald Trump pushes for a single national framework to govern AI, U.S. states are not waiting. Across the country, legislators are drafting and passing laws that directly shape how AI can be used in health care settings, from insurance claim reviews to mental health chatbots. What looks like a technical policy debate is quickly becoming a power struggle over who gets to decide the future of medical AI, and how much risk society is willing to tolerate along the way.

Summary: States Move First as Federal AI Rules Lag Behind

As of mid-October, lawmakers in 47 states introduced more than 250 bills addressing artificial intelligence in health care, according to tracking by Manatt, Phelps & Phillips. Thirty-three of those measures have already become law across 21 states, signaling an unusually fast pace of regulation for a technology still evolving in real time. Many of these efforts are bipartisan, with traditionally red and blue states borrowing from each other’s proposals rather than following strict ideological lines.

A significant portion of the legislation focuses on AI-powered chatbots used in medical and mental health settings. Several states have acted after research and real-world cases suggested that chatbots can produce unsafe or misleading responses, particularly for teenagers seeking mental health support. Illinois, for example, enacted a law that bans apps or services from using AI chatbots to deliver mental health or therapeutic decision-making. Other states require chatbots to clearly disclose that they are not human, reliably detect crisis situations, and avoid generating harmful advice.

Insurance practices are another major target. States are scrutinizing how insurers and managed care plans use AI to review treatments, deny claims, or predict patient costs. Some bills require companies to report how often AI-assisted decisions result in contested claim denials, aiming to expose whether automated systems are unfairly restricting access to care. Transparency and anti-discrimination measures are also spreading, particularly for so-called high-risk AI systems that influence eligibility decisions in health care, insurance, housing, or education.

Clinical settings have not been ignored. States like Texas, Nevada, and Oregon have passed laws to address potential bias, misuse of sensitive health data, and overreliance on automated tools as providers increasingly deploy AI to reduce administrative burdens. At the same time, Colorado recently delayed implementation of a sweeping transparency law after facing heavy lobbying from the tech industry and concerns about compliance costs.

These state-level actions are colliding with Trump’s renewed push for federal control. A recent executive order directs the attorney general to form a task force aimed at challenging what the administration views as burdensome state AI regulations. It also calls on Congress to draft legislative recommendations for a nationwide AI framework that could override state laws. This tension has already surfaced in disputes between the White House and AI companies like Anthropic, which publicly supports state transparency initiatives while also advocating for a federal standard.

The debate is expanding beyond jurisdiction. A bipartisan group of 42 state attorneys general recently pressured major AI companies, including OpenAI, Meta, Google, and Microsoft, to introduce safeguards protecting children and vulnerable users from emotional manipulation and other harms linked to generative AI. Meanwhile, industry leaders warn that a fragmented legal landscape could slow innovation, create confusing reporting requirements, and hinge on vague distinctions between “machine learning” and “artificial intelligence” itself.

What Undercode Say:

The rapid rise of state-level AI regulation in health care reflects a deeper truth that Washington often avoids: health technology does not fail in theory, it fails in practice. States are reacting to concrete risks, chatbots offering unsafe mental health advice, algorithms denying care, and opaque systems influencing life-altering medical decisions. From that perspective, state lawmakers are not overreaching, they are filling a vacuum.

However, the emerging patchwork is undeniably messy. AI does not respect state borders, especially when cloud-based models serve patients across multiple jurisdictions. A mental health chatbot developed in California can be used by a teenager in Texas within seconds. When each state defines AI risk differently, companies are forced to either overcomply or limit access, both of which can reduce innovation and availability of care tools.

Trump’s push for a federal framework makes strategic sense, but timing matters. A national standard imposed too early risks locking in weak protections or outdated definitions of AI. Imposed too late, it becomes a blunt instrument overriding more nuanced state safeguards. The current clash suggests that the federal government is reacting to state momentum rather than leading it.

Another critical issue is definitional ambiguity. Many bills hinge on whether an AI system is involved in making a “consequential decision,” a term that remains legally fuzzy. Insurers and providers may argue that AI only supports human judgment, while regulators view the same system as functionally decisive. This gray area will likely drive litigation and slow enforcement.

The mental health focus deserves special attention. States are right to be alarmed by AI companions and chatbots aimed at minors. Emotional dependency, misinformation, and failure to detect crises are not hypothetical risks, they are already documented. The bipartisan pressure on major AI firms shows that child safety could become the rare issue where federal and state interests align.

Ultimately, the conflict is less about innovation versus regulation and more about trust. Health care is one of the few domains where algorithmic mistakes can cause irreversible harm. States are signaling that transparency, accountability, and human oversight are non-negotiable. The federal government now faces a choice: harmonize these efforts into a credible national framework, or attempt to suppress them and risk public backlash when AI systems inevitably fail.

Fact Checker Results

✅ State-level AI health care legislation has expanded rapidly across nearly all U.S. states.
✅ Multiple laws now regulate medical chatbots, insurance AI use, and transparency obligations.
❌ No unified federal AI framework currently exists that preempts state health care AI laws.

Prediction

📊 Over the next two years, federal lawmakers will introduce a national AI health care framework, but it will likely preserve carve-outs for mental health and child protection.
📊 States will continue to act as testing grounds, shaping the final federal standards through pressure rather than compliance.