Guest commentary: Artificial intelligence is inflating medical bills

Just when the healthcare system seemed like it could not get any more confusing, artificial intelligence (AI) has brought a new set of problems.

Medical debt, or money people owe for healthcare, seems likely to become even harder to deal with because of these problems.

Look at two study reports published this year:

A study released on Oct. 6 by the Robert Wood Johnson Foundation found that more than one in four adults, or 27%, said they or someone in their family had medical debt in December 2025. Another report, released by J.P. Morgan on Aug. 17, found early successes in using AI to manage healthcare payments, decide which patients need care first, get insurance approval before treatment, and handle medical bills.

Looking at these two reports together, it seems clear who is most likely to lose money and who is most likely to make money as AI becomes more common in healthcare.

Companies such as Cedar are already changing how healthcare providers use technology to send bills and arrange payments for patients. Cedar is a healthcare billing platform powered by AI. It offers what the company calls a payment experience that is made to fit each patient's needs.

A report released by BlueCross Blue Shield in March also found many cases of what it called "AI upcoding." The report looked at insurance claims from tens of thousands of hospital stays for childbirth and maternity care. It found that this practice added $22 million to the cost of maternity hospital stays in one year.

AI upcoding happens when AI makes a patient's medical diagnosis appear more complicated or serious. This allows a hospital to charge insurance companies more money.

AI tools are being promoted as technology that can greatly improve healthcare. But these tools are also making it easier to push healthcare costs higher than they should be.

This creates a new problem for hospitals, especially nonprofit hospitals, that want to use technology to handle patient bills automatically and make their billing systems more efficient.

More AI technology is being used every day. It is like squeezing toothpaste out of a tube: once it is out, putting it back is very difficult. Because AI is already spreading through healthcare, even suggesting a ban on its use in healthcare organizations no longer seems realistic.

A better option would be to create clear rules about how AI is allowed to be used.

First, hospitals need to check whether patients qualify for financial help before using automatic billing systems. Right now, these systems seem to follow a "collect the money first, ask questions later" approach.

Tradeoffs, a nonprofit news organization that reports on healthcare systems, investigated the rules used by 166 nonprofit hospitals to decide whether patients likely qualify for financial help. It found that in Texas, 44% of hospitals often sent patients bills before checking whether they qualified for free or lower-cost care.

Using AI more often to send bills and create payment plans that are different for each patient is making an existing problem even worse.

Making payment plans fit individual patients is not necessarily unfair or discriminatory. But whether AI billing systems will clearly show people how they make decisions is another question. A lack of openness could be an even more serious problem.

Texas already has a law about clear medical bills. Section 185.002 of the Texas Health and Safety Code requires healthcare providers to give patients detailed bills. These bills must list every service and medical supply provided during a visit and explain each one in plain, easy-to-understand language.

The real improvement AI could bring to healthcare would be for hospitals to use computer programs that make patients' legal protections stronger. Hospitals should not use AI to find ways around the law or avoid taking responsibility by blaming decisions on a computer program.