The data: Consumers are still worried about medical AI tools operating without human oversight and the accuracy of AI diagnoses, but those fears are easing slightly as data privacy concerns rise, according to a September Axios-Ipsos survey of 1,122 US adults.

• Consumers are less worried about medical AI operating without human oversight: 62% cited lack of clinician oversight as a top concern, down from 68% in June and 69% in 2024.
• But more concerned about their data privacy: 50% cited medical data privacy and security as a top concern, up from 44% in June and 38% in 2024.
• Concerns over diagnostic precision have eased since June but remain elevated compared with 2024: 57% are concerned about AI’s diagnostic accuracy, down from 63% earlier this year but up from 54% in 2024.

Why it matters: Falling concern over human oversight suggests consumers are growing more comfortable with automated healthcare as they use AI themselves for health guidance. In fact, more people are acting on AI guidance for healthcare decision-making, sometimes without a physician’s input. Meanwhile, AI tools handle clinical tasks previously performed by physicians, such as prescribing certain medications and even diagnosing patients.

More personalized AI experiences in healthcare also heighten data privacy concerns. AI tools are moving beyond answering general health questions to analyzing medical records and other sensitive information. OpenAI’s Health in ChatGPT consumer product and a growing number of health apps and wearables encourage users to share medical records and health data for personalized AI guidance. That raises questions about how information is stored, shared, and protected: 67% of US adults say they’re uncomfortable sharing health records with AI platforms, per a March Cloaked report.

Implications for healthcare organizations and AI companies: Consumers can separate AI's capabilities from the companies deploying it. They turn to AI for useful health guidance but remain skeptical that tech firms, and even healthcare organizations, will adequately safeguard their medical information. Human oversight and the tech’s clinical accuracy will remain crucial as AI takes on higher-stakes patient care tasks, but data privacy could become the deciding factor in whether consumers trust AI-powered healthcare.

AI companies and healthcare providers that use AI for patient care should make data privacy central to their value proposition. That means clearly explaining how health information is used and whether third parties can access it, and offering a simple way to revoke access or delete data.