🔎︎
Healthcare workers across the country are using unapproved artificial intelligence tools, raising significant safety and liability concerns for health systems and payers. A survey published by Wolters Kluwer Health reveals that more than 40% of medical workers and administrators know colleagues who use unauthorized products, while nearly 20% admitted to using shadow AI tools themselves.
The widespread use of unapproved AI products represents a critical governance gap with implications for clinical quality, patient safety, and organizational risk management. Healthcare workers reported turning to unauthorized tools primarily because approved alternatives were unavailable, too slow, or lacked necessary functionality. More than half of administrators and 45% of care providers reported using shadow AI to speed up their work. Almost 40% of administrators and 27% of providers said they used it because approved products didn't exist or didn't work as well.
Dr. Peter Bonis, chief medical officer at Wolters Kluwer, emphasized that unapproved tools bypass essential vetting processes for safety, efficacy, and risk management. The concern extends beyond workflow efficiency to fundamental questions about accuracy and clinical decision support, particularly since AI tools can generate misleading or inaccurate information that may not be adequately intercepted at the point of care.
Patient safety concerns ranked highest among survey respondents, with about one-quarter identifying it as their top AI-related worry. For managed care organizations evaluating provider networks and clinical performance, the use of shadow AI introduces uncontrolled variables that affect care quality and consistency across contracted providers.
The cybersecurity implications also warrant attention from payers and health plans. Because shadow AI operates without institutional oversight, IT teams lack visibility into how these tools access, process, or transmit protected health information. Healthcare cybersecurity already faces substantial challenges given the sector's valuable data stores and high-stakes care delivery, making unauthorized AI adoption particularly concerning for organizations managing population health data.
Policy awareness gaps compound the governance challenges. 29% of providers said they knew about their organization's main AI policies, but only 17% of administrators said they did. This disconnect suggests that even when AI governance frameworks exist, they may not effectively reach the clinical workforce when making daily technology decisions.
More than 500 people from hospitals and health systems across the country took part in the survey. Over 25% of providers and 10% of administrators said they use shadow AI because they are curious or want to try new things, suggesting that some unauthorized use stems from a genuine interest in innovation rather than a lack of approved tools.
