Older Adults Trust the Doctor in the Room, Not the AI Health Tool

Artificial Intelligence (AI) is increasingly showing up in everyday life, including in health care. For many older adults, health care is complex, confusing, and emotionally charged. AI and Generative AI, at their best, can help people navigate their health care and simplify access to health information on symptoms, medications, and insurance statements.

Generative AI is already becoming a quiet health companion, as a recent AARP online survey found that 2 in 5 adults over age 50 who use AI (41%) have already asked an AI-powered tool a health-related question. In the next 12 months, even more AI users (3 in 5, 62%) say they are likely to ask an AI-powered tool a health-related question.

But there is a critical gap between use and trust. While many users of AI-powered tools have asked health-related questions, few have shared, or are likely to share, personal health documents such as test results or insurance statements (have shared: 11%; likely to share in the future: 23%). Even fewer have directly connected their medical records or wellness apps to an AI-powered tool for ongoing monitoring and insights (have shared: 4%; likely to share in the future: 21%).

In fact, 69% of older adults, both AI users and nonusers, do not feel comfortable sharing their personal health information or medical documents with a health-specific generative AI-powered tool, chatbot, or agent.

Why? Because they trust the doctor in the room the most. When asked about reasons for not feeling completely comfortable sharing personal health information, respondents’ preference for having a doctor explain their health rather than an AI tool was the top reason (63%). Other reasons included:

• Concerns about information being shared or sold without consent (61%).
• Uncertainty about how information would be used by the AI system (58%).
• Privacy and security concerns regarding their health data (58%).
• Lack of regulations or oversight governing AI companies (57%).
• Lack of trust or understanding regarding accountability if the AI provides incorrect or harmful guidance (57%).

These concerns are valid. Over half of older adults (57%) are unsure who will have access to their personal health information in the future once it is entered into a generative AI tool. In fact, 34% of older adults believe that if an AI tool states it is “HIPAA compliant,” any entered information will be safeguarded by regulations that ensure confidentiality, and 6% believe they are the only ones who will have access to the information they share or upload. Both assumptions are incorrect.

In reality, most AI-powered tools are not required to meet the same privacy and data protection standards as health care platforms. Users may unknowingly give up rights to their data, sometimes permanently. One example in the real world illustrates this. When a user of an AI-powered tool changed his data-sharing settings, everything he had uploaded disappeared for good, revealing how little control and clarity consumers actually have.

Additionally, the inaccuracy of AI-powered health tools may have dangerous consequences. A recent independent evaluation published in the journal Nature Medicine found that ChatGPT Health “under-triaged” or failed to recognize the seriousness of more than half (51.6%) of the medical emergency scenarios tested. Instead of telling patients to go to the hospital immediately, the AI suggested they stay home or book a doctor appointment for life-threatening conditions like respiratory failure and diabetic crises. In one simulation, the AI failed to realize a woman was suffocating and instead directed her to a future appointment she would not have survived to attend.

Older adults are willing to use AI to help them navigate health care, but their openness comes with a condition: They need to trust it. Specifically, they expect AI-powered tools to protect their personal health information in the same way other health care services do. To foster responsible consumer adoption, companies and other stakeholders should provide plain-language education about:

• What AI-powered tools are and are not legally required to protect.
• Who owns the data, how long it is kept, and how users can request its deletion.
• What safeguards are in place to bring AI tools in line with the privacy standards people already expect from health care.

Methodology

Interviews were conducted between February 12 and 16, 2026, among 1,031 U.S. adults age 50-plus in the Foresight 50+ Omnibus. Funded and operated by NORC at the University of Chicago, Foresight 50+ is a probability-based panel designed to be representative of the U.S. household population age 50 or older. Interviews were conducted online and via phone. All data are weighted by age, sex, education, race/ethnicity, region, and AARP membership.

For more information, contact Brittne Kakulla at bkakulla@aarp.org. For media inquiries, contact External Relations at media@aarp.org.