According to Gizmodo, Utah has launched a first-in-the-nation pilot program allowing an AI system to renew 190 commonly prescribed medications for chronic conditions. The program, developed with health-tech startup Doctronic, excludes drugs with abuse potential like pain meds and ADHD treatments. It will initially cost patients $4 per renewal, with plans to shift to insurance coverage or an annual fee. Doctronic’s AI matched human clinician decisions 99.2% of the time and will be held to the same malpractice liability as a doctor. The company, which launched in 2023, already offers an AI chatbot for free medical consultations. Meanwhile, an OpenAI report claims 46% of American nurses use AI tools weekly.
The Speed vs. Safety Tradeoff
Here’s the thing: the promise is huge. Easing the burden on overworked doctors and getting people their maintenance meds faster? That’s a win. The logic behind Doctronic is all about removing barriers, and 70% of healthcare AI chats happening after hours shows a real need. But this is medicine. A 99.2% match rate sounds impressive until you remember that the 0.8% could involve a fatal drug interaction the AI missed. We’re not talking about an AI recommending a movie you might not like. This is serious. And we already know these systems can be tricked or have blind spots.
The Bias Problem Isn’t Theoretical
This is my biggest worry. A recent Financial Times report highlighted that some medical AI tools downplay women’s symptoms and stereotype races. So what happens when that biased algorithm is the one deciding if your refill request is valid? The state says the AI will be held to a doctor’s standard of care legally, but that feels like closing the barn door after the horse has bolted. If a biased system denies or incorrectly approves a medication, the damage is done. Holding a company liable doesn’t un-harm the patient. It’s a massive, unproven experiment in real-time.
A Legal and Regulatory Gray Area
The legality here is fascinating, and honestly, a bit of a loophole. States regulate medical practice, so Utah can basically say, “This AI is practicing under our rules.” But wait—the FDA regulates medical devices, and an AI that prescribes sure seems like one. The FDA is actually reevaluating how it oversees AI in health right now. So Utah is sprinting ahead while the federal referee is still figuring out the rulebook. Doctronic is already in talks to expand to Texas, Arizona, and Missouri. This could spread fast, creating a patchwork of state-level approvals before any cohesive national safety framework exists.
Where Do We Go From Here?
Look, AI in healthcare is inevitable. The genie is out of the bottle. But moving from assisting doctors to fully replacing their judgment for prescription renewals is a massive leap. The Politico report makes it clear this is starting with a limited set of low-risk drugs, which is smart. The pilot needs to be watched like a hawk. Every error, every near-miss, has to be public data. Because if the goal is truly better access, we can’t sacrifice safety for speed. And if we’re going to hand this responsibility to algorithms, we need to be 100% certain they see all patients—regardless of gender, race, or background—with perfectly clear eyes. I’m not convinced we’re there yet. Are you?
