Asking an AI chatbot whether your blood pressure monitor is actually trustworthy could leave you with a wrong answer. New research presented at the American Heart Association’s Hypertension Scientific Sessions 2026 found that popular AI search tools can correctly identify validated monitors only 63-91% of the time, depending on which AI chatbot you use.
AI can identify validated home BP monitors only 2/3 of time
For anyone unfamiliar with the term, “validation” means independent testing has confirmed a device consistently produces accurate readings. A home blood pressure monitor counts as validated once independent, third-party testing confirms it consistently gives accurate readings, and three official registries track which devices pass that bar.
Researchers tested 324 devices, split between validated and non-validated monitors, by asking Google Gemini, Microsoft Copilot, ChatGPT, and Perplexity three different ways to identify validation status.
Test results showed that Gemini performed best, landing between 86% and 91% accuracy, while the other three tools ranged from just 63 – 83% percent. Oddly enough, every single tool struggled more with correctly confirming a validated device than flagging an unvalidated one, even when that device was clearly listed on an official registry the AI had access to.
Consistency is also a problem
When researchers asked the same AI tool the exact same question again, on a different day or a different computer, it sometimes gave a completely different answer for the same device. Study author Dr. Anna Soriano said most tools performed only slightly better than a coin flip, and even Gemini often missed information that was easy to locate.
The inconsistency matters because a wrongly validated device could lead to inaccurate readings and, down the line, the wrong diagnosis or treatment decision. The findings are a reminder that AI still needs real caution in clinical decision making.
Why should you care about an AI’s response?
A validated monitor is crucial because inaccurate readings can affect medical decisions. Researchers warned that people could trust an incorrect AI answer and choose a device that fails clinical standards.
For now, researchers recommend skipping AI chatbots entirely and checking a monitor’s status directly through a recognized registry like validatebp.org instead.
RELATED COVERAGE: