What Trump’s executive order on restricting state AI laws means for healthcare

Congress doesn’t want it.  State leaders don’t want it.  Consumers don’t want it.  But the leader of the party of states’-rights-only-when-it-suits-their-agenda seems set on trying to make it happen.

Last week, the Trump Administration issued an Executive Order establishing a task force to evaluate and challenge state regulations related to artificial intelligence (AI) that the White House doesn’t like, essentially telling states that they won’t be allowed to make their own rules anymore unless they pass muster.

Criteria for scrutiny include laws that “unconstitutionally regulate interstate commerce, are preempted by existing Federal regulations, or are otherwise unlawful in the Attorney General’s judgment,” such as those that “require AI models to alter their truthful outputs, or that may compel AI developers or deployers to disclose or report information in a manner that would violate the First Amendment or any other provision of the Constitution.”

If the order sticks, states that fail the test would be deemed ineligible for funding from the $42.5 billion Broadband Equity Access and Deployment (BEAD) program, which is designed to expand broadband internet access to underserved and rural regions.

While the EO does say that it isn’t supposed to preempt “lawful” state-level regulations, it is still intended to have a deeply chilling effect on states that don’t want to risk important infrastructure funding and get tangled up in expensive, protracted legal battles with the current Administration around local policymaking.

The order immediately sparked pushback from bipartisan policy experts, who questioned whether the tactic was legal, as well as prominent members of the Republican party, who decried the Administration’s multiple efforts to sneak similar provisions into big package legislation.

Previous tries have included a failed attempt to implement a 10-year moratorium on lawmaking this summer, in which the Senate voted 99-1 against including the provision, and another non-starter attempt in the pending defense bill this month.

Why are state-level AI laws a target for the federal government?

There are, in fact, some legitimate issues to think about when it comes to state-level versus federal-level regulation of AI.  It’s true that conflicting or unclear state laws can cause confusion for developers and users, especially those that market their products across state lines or have user bases in multiple regions.  And it’s plausible to assume that an uncoordinated patchwork of policies could make it more difficult to roll out technologies in a speedy and compliant manner.

But here’s the rub: the federal government has been stubbornly averse to adequately providing the structured, unified consumer protections required for states to confidently ease off doing things on their own.

For example, on Trump’s first day in office, he garroted President Joe Biden’s 2023 executive order on AI ethics and slashed other Biden-era actions on federal regulation of AI development and deployment.

He has also reduced regulations around building data centers, tried to manipulate the design of AI algorithms by twisting the real scientific issue of algorithmic bias into something that fits his “anti-woke” agenda, and given Big Tech leaders unprecedented influence and access at the federal level – none of which actively contribute to protecting consumers in real life while creating a meaningful, nationalized framework for AI innovation.

Even the federal AI action plan released in July stresses deregulation as the pathway to innovation.

There’s question that unnecessary or duplicative barriers should be removed.  But taking the concept too far, while simultaneously restricting states from closing gaps they observe in their own communities, will result in a dangerous policy vacuum for a category of technologies that are radically altering the very nature of our social fabric.

With hundreds of billions of dollars on the table, and some unscrupulous AI developers already eager to exploit the free-for-all attitude of the current White House, there’s a good chance that consumers are the ones who are going to get the short end of this particular stick.

The public already knows this, and they’re dead set against letting AI companies run amok.

In one recent Gallup poll, 80% of respondents (including 79% of Republicans) said they want strong rules around AI safety and security, even if it means developing technology at a slower pace.  And more than half (54%) want to see the federal government take charge of policymaking to ensure some semblance of clarity and consistency as AI develops, which is about as much of a mandate as one can expect these days.

What are the implications for the healthcare industry?

Healthcare is particularly vulnerable to the ill effects of an order preventing states from monitoring and managing what’s happening in their own backyards.  There are life and death risks to unregulated AI, both in the clinical and administrative spheres.

Some of these issues are already surfacing, particularly in the payer sector, where lawsuits about improper AI usage for claims determinations have grabbed headlines in recent years.

In the absence of federal activity, states have taken up the challenge of regulating how payers use AI tools to make decisions about coverage and eligibility, with several laws enacted this year preventing health plans from relying on AI alone to approve or deny services.

States like California and Illinois have also started to address the use of AI in the clinical arena by requiring disclosure of generative AI material and banning AI from acting independently around mental health care.  Meanwhile, many other states have started to pass laws demanding greater transparency, more robust auditing, and enhanced guardrails around autonomous AI in the clinical setting.

It’s a good start, and it needs to continue.  However, if the new EO renders states unable to make even these initial strides toward creating a regulatory framework for AI in healthcare, it will be up to individual healthcare organizations to devise, deploy, and adhere to rigorous rules around how AI is applied to clinical or operational use cases.

And to be honest, they haven’t really gotten the knack of doing so yet.  AI governance is already hugely challenging, even to large, digitally mature organizations with the experience and resources to invest in governance activities.

Instead, organizations are likely to find themselves wasting resources on reinventing the wheel of governance while burning time and money on deeply investigating the trustworthiness, security, and reliability of vendors whose products and offerings are not beholden to any external regulatory criteria.  For those organizations that don’t even have a firm understanding of what factors and flags they should be looking for, an already onerous task will become nearly impossible to complete correctly.

The real risks are that without preemptive regulations with teeth behind them, patients will die before problems with vendors who have cut corners start to come to light, which is the worst case scenario for all involved.

Industry-level frameworks and voluntary commitments can be valuable, but they only hold so much weight.  There is an urgent need for more regulation around AI, not less, if the industry is to safely, efficiently, and effectively adopt AI tools that hold so much promise when utilized appropriately.

What happens now?

It’s likely that the Executive Order will face legal challenges, which may delay or prevent implementation from going forward.  State leaders from both parties are unlikely to allow this EO to set a precedent for federal power to override their lawmaking ability in other areas, so chances are this particular decree won’t be around for too long.

However, the broader trend of aggressive deregulation at this pivotal juncture in the AI maturity curve could be even more damaging than any individual proclamation.  Healthcare organizations will need to keep a sharp eye on the regulatory environment to see which actions are helping or hurting their ability to adopt technology safely while ensuring that patients and organizations both receive adequate protection.

Taking an active role in advocacy at the industry, state, and federal levels will be essential for designing a regulatory environment that appropriately balances helpful guardrails with critical opportunities for innovation, creating a space for AI to thrive and show its full potential for improving the delivery of patient care.