AI Applications Lead ECRI's 2025 Health Technology Hazards List
Patient safety organization ECRI ranks artificial intelligence-enabled applications as the top health technology hazard for 2025 due to risks of bias and misleading outputs. The annual list also highlights unmet home care tech needs, cybersecurity threats, and fraudulent medical devices.
ECRI ranks artificial intelligence-enabled applications as the number one health technology hazard for 2025. The patient safety organization warns that biases in AI training data lead to disparate health outcomes and inappropriate responses. Placing too much trust in these models results in inappropriate care decisions, especially when the technology generates false or misleading outputs.
The nonprofit elevates AI to the top spot after ranking insufficient AI governance fifth in 2024. ECRI emphasizes that AI models struggle to adapt to new conditions, experience changes in performance over time, and often yield disappointing results when organizations set unrealistic expectations. Healthcare facilities frequently fail to define clear goals, provide adequate oversight, or properly prepare their data for these systems.
Despite these significant warnings, ECRI acknowledges the potential value of AI as a tool to assist clinicians and increase the efficiency of medical diagnoses. The organization insists that human decision-making must remain at the core of the care process. Other major hazards on the 2025 list include the unmet technology support needs of home care patients, cybersecurity threats, and substandard or fraudulent medical devices.