The Possible Demise of a Federal Program Isn’t a Rejection of AI in Healthcare

Thanks to AI, U.S. healthcare stands at a crossroads.

On the one hand, healthcare organizations are racing to transform the way care is delivered. Investors have realized that AI has the potential to upend the market and bring massive gains. In my own sector – healthcare administration – this takes the form of automated paperwork. In others, like radiology, LLMs are identifying and diagnosing medical conditions.

Massive funding is being pumped into AI in healthcare. Just last year, U.S. healthcare AI companies secured more than $18 billion  in venture investment – up nearly 10% from 2024. This rapid growth is projected to continue and will change much of what we know about healthcare provision.

On the flip side is the governance challenge. Regulators are scrambling to understand AI and legislate it effectively. We all know that the state has a responsibility to its citizens. But if the industry changes quickly, it can be hard for governments to keep up.

Nowhere do we see this dynamic more than with health insurance. In the U.S., hospitals and doctors must frequently get prior authorization from the patient’s insurance company before delivering care. Insurers reserve the right to deny treatment they deem not medically necessary.

AI healthtech promises to increase the entire sector’s efficiency, but it must be carefully managed. The CMS (Centers for Medicare & Medicaid Services) estimates that 25% of all costs in the industry are administrative waste. AI can reduce this burden, benefiting the public. Nevertheless, we need the right principles for success. 

Enter WISeR. Introduced by CMS at the start of this year, the Wasteful and Inappropriate Services Reduction (WISeR) program is one of the largest pilots in AI healthtech. Using AI models to summarize evidence, qualified clinicians are attempting to adjudicate claims in the traditional Medicare program far more quickly. The experiment is currently live in six states.

In June, the House Appropriations Committee voted unanimously to block WISeR’s funding. The budget still needs to clear Congress, and the pilot remains active. But its future beyond September 30th remains unclear.

What I want to say here is that – whatever we think of the WISeR vote – the pilot’s fundamental principle remains firm. “AI flags, clinician affirms” continues to provide the foundation for good governance in AI healthtech.

Lawmakers called for better processes

While it’s easy to read this decision as a rejection of AI in healthcare, that would be a mistake. The vote to defund WISeR was made amid concerns over implementation and incentive structures – not because lawmakers opposed AI adoption in clinical review.

The main reason the House committee gave for its decision was that it introduced prior authorization to traditional Medicare. The 13 medical services selected for the pilot had previously only required a doctor’s consent, not insurer authorization. Lawmakers therefore rejected WISeR for what they deemed increasing administrative complexity and changing federal incentives.

This is not a criticism of AI healthtech. Instead, it’s a criticism of how the process has been designed.

“AI flags, clinician affirms” is still healthtech’s foundation

While the fate of WISeR is now ambiguous, its central tenet is not. The principle that “AI flags, a clinician affirms” remains the foundation of modern healthtech. 

This principle is central because of AI’s status as an emergent system. Unlike other forms of automation, LLMs have the potential to make decisions that their creators didn’t intend. This is by design. Emergence allows AI to behave flexibly and creatively, but it also brings risk. Regardless of how refined these systems become, regulators will always need a “human in the loop” to catch random decisions and ensure patient safety.

In practice, this means that a qualified clinician retains all legal and ethical responsibility when using AI. LLMs gather and summarize evidence at machine-speed, highlighting important findings. Clinicians review the data, concisely presented, and use it to inform their judgement. This process promises to increase efficiency while retaining maximum accuracy. Clinical judgment is strengthened because humans remain central to the process, not despite their involvement, but because of it.

At present, we can see this consensus building at all levels of the healthcare industry.

Last year, the nation’s largest health insurers pledged their commitment to improving the prior authorization process. This includes increasing transparency and accelerating decision timelines, with an emphasis on immediate approvals for certain urgent or clinically standard treatments, which has led to AI becoming a virtual necessity for many insurers. The leading industry association for independent review organizations also outlined its position on the use of AI.

The state floor of regulation is also rising. Already, states possess relevant consumer protection laws, which are now being adapted to AI. So far in 2026, 43 states have introduced 240 healthcare AI bills. Of these, nine prohibit using AI as the sole determinant of claims denial.

If the WISeR program is ultimately halted, it wouldn’t affect the consensus that “AI flags, clinician affirms”. With it, we have everything we need for success in the industry. We just need to refine the structure of AI implementation.

AI adoption is non-negotiable

Given AI’s efficiency potential, healthcare entities that are slow to adopt the technology will be quickly outcompeted. 

The lesson to take from WISeR isn’t that we can’t have AI in healthcare. Far from it. Instead, we need to recognize the potential LLMs have to improve efficiency in healthcare. For this change to occur, qualified professionals must remain central. The consensus of “AI flags, clinician affirms” provides the base for transition, already setting the standard across much of the industry.

AI promises significant improvement in efficiency and access to care, but it cannot operate without human review. WISeR shows that we can, and must, remain in the loop. Not as an add-on, but as a necessity.