One in Three Americans Now Asks AI Chatbots for Health Advice—With Mixed Results

A third of U.S. adults (32%) now use AI chatbots to get advice about physical or mental health, putting them roughly on equal footing with social media as a place people go for health information, according to KFF tracking poll data published in March 2026. But here's the catch: people who use these chatbots regularly are buying into false claims about vaccines at nearly the same rate as those who rely on social media.

The numbers, from a KFF survey conducted May 7–31, 2026 and released in June, are striking in their similarity. Among people who use AI chatbots at least weekly for health information, 35% incorrectly believe vaccine myths. For weekly social media users seeking health information, the figure is 37%. The same survey found something else: people without a trusted doctor or healthcare provider are more likely to hold these false beliefs. When someone lacks access to a real provider and is getting health information from digital channels, that combination makes misinformation stick harder.

Why This Matters

The overlap between AI and social media misinformation rates deserves real attention. Social media's role in spreading health falsehoods has been examined closely since COVID-19, and platforms have faced pressure from regulators and public health officials. AI chatbots are newer to this conversation. Early talk around them suggested that large language models—which pull from medical databases and published clinical research—would be better at avoiding misinformation than the unmoderated chaos of social feeds. The KFF data don't support that optimism.

Trust in these tools is also low. Only about one in three adults (32%) say they would be comfortable with an online health tool using AI to access their personal medical records, per KFF polling from January 2026. People are already using these tools at high rates, but they don't yet trust them with sensitive health data—a familiar pattern from earlier health technologies like patient portals and at-home genetic testing, where usage always outpaces confidence.

For doctors, hospitals, and public health officials, the message is clear: AI chatbots are no longer niche. A third of the public already uses them for health advice, at a scale that requires the same careful attention that shifted toward social media about a decade ago. The fact that heavy users of both AI and social media believe false vaccine claims at similar rates suggests the problem isn't unique to how either platform is designed. It may be rooted in how people seek out information in the first place, or in the gap when someone doesn't have a doctor they trust.

That last finding—about the absence of a trusted provider—might be the most useful takeaway for public health planners. Belief in misinformation appears to track with whether someone has access to a healthcare provider, not just which platform they're using. Focusing only on filtering bad content or improving AI accuracy may miss a bigger structural problem: people don't have doctors in the first place. The platform is a symptom. The missing access is the disease.

One remaining question the data don't answer: Are AI chatbot users a completely different group from social media health information users, or are they largely the same people bouncing between both? Understanding that overlap matters for knowing how many people are actually exposed to myths, and for designing outreach that doesn't overcount. KFF's tracking surveys in the months ahead will likely be the first place to watch as these numbers shift.