As artificial intelligence becomes embedded in everyday health searches, new research presented at the American Heart Association's Hypertension Scientific Sessions 2026 reveals a troubling gap: three of four leading AI-powered search tools accurately identified home blood pressure monitors that met validated clinical standards only 63% to 83% of the time. The findings, announced via NEWMEDIAWIRE, carry significant implications for the more than 125 million U.S. adults with high blood pressure, a condition that is the number one risk factor for cardiovascular disease.
According to the American Heart Association's 2026 Heart Disease and Stroke Statistics Update, about 47% of U.S. adults have high blood pressure, yet only about 1 in 4 of those adults have their blood pressure within the target range of less than 120/80 mm Hg. The The 2025 American Heart Association Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults recommends using a home blood pressure monitor that has been validated for accuracy, and people can visit www.validatebp.org to check device listings.
The study tested 324 home blood pressure monitors—145 validated and 179 not validated—using four AI search tools: Google Gemini, Microsoft Copilot, ChatGPT and Perplexity. Google Gemini performed best, answering correctly 86% to 91% of the time, while the other three tools ranged from about 63% to 83% accuracy. All four tools were less accurate at identifying validated devices compared with unvalidated ones, and when researchers retested a subset of devices on different days and computers, the tools often changed their answers.
"We found that most AI tools performed only slightly better than if you had flipped a coin for each question," said Anna Soriano, M.D., a third-year internal medicine resident at the University of Montreal and the study's presenting author. "People may unknowingly think a device is validated based on the AI tool's inaccurate responses. Using that device may result in inaccurate blood pressure readings, which could lead to an inappropriate diagnosis or treatment decisions."
The authors could not determine exactly why AI tools struggled, but Soriano noted it is "surprising, and almost counterintuitive" that AI had difficulty identifying validated devices, since those are the ones with clear listings on official registries. Keith C. Ferdinand, M.D., FAHA, an American Heart Association volunteer expert and vice chair of the Association's 2025 High Blood Pressure Guideline, said the results should remind clinicians and the public that "the use of AI for clinical decision-making requires caution." He added that home blood pressure devices need to be both validated and accurate, and that with proper technique, home readings are a valuable component of integrated treatment plans.
The study's limitations include that AI technology continues to improve, and private browsing sessions could not fully prevent language model training. The findings are considered preliminary until published as a full manuscript. Additional resources include the Cuffless blood pressure technologies in wearable devices show promise to transform care release and guidance on Home Blood Pressure Monitoring. The abstracts will be published in the Hypertension Scientific Sessions 2026 Supplement of the Hypertension journal. More information is available at heart.org and stroke.org.
