A third of U.S. adults (32%) now turn to AI chatbots for physical or mental health advice, putting them roughly on par with social media as a health information channel, according to KFF tracking poll data published in March 2026. A separate KFF poll — fielded May 7–31, 2026 and released on June 30 — adds a sharper edge to that figure: frequent AI chatbot users endorse vaccine myths at nearly the same rate as frequent social media users.

The June 2026 numbers are close to identical across the two channels. Among adults who use AI chatbots at least weekly for health information, 35% incorrectly endorse vaccine myths. Among those who rely on social media at least weekly for the same purpose, the figure is 37%. KFF's June 2026 Tracking Poll on Health Information and Trust also finds that people without a trusted healthcare provider are independently more likely to hold those false beliefs — a compounding vulnerability when provider access and digital health-seeking behavior overlap.

The convergence of AI and social media misinformation rates is worth sitting with. Social media's role in health misinformation has been litigated extensively since the COVID-19 pandemic, with platforms under sustained regulatory and public-health pressure. AI chatbots are newer to that conversation, and the assumption embedded in much of the early discourse around them was that large language models — drawing on indexed clinical literature and, in some deployments, proprietary medical databases — would outperform the unmoderated churn of social feeds. The KFF data, while not designed to establish causation, don't support that assumption.

The trust picture is equally constrained. Only about one in three adults (32%) say they would trust an online health tool that uses AI to access their personal medical records, per KFF polling from January 2026. Adoption is running ahead of trust — a structural tension familiar from earlier waves of health technology, from patient portals to direct-to-consumer genetic testing, where uptake consistently preceded public confidence in data handling and clinical reliability.

For health policy professionals and clinicians, the operational implication is direct. AI chatbots are not a fringe information source; they are already a primary channel for a third of the public, at a scale that warrants the same institutional attention that shifted toward social media a decade ago. The fact that frequent users of both channels endorse vaccine myths at statistically similar rates suggests the problem is not specific to platform architecture — it may be more closely tied to the underlying information-seeking behavior, the absence of a trusted clinical relationship, or both.

That last thread — the trusted provider finding — is arguably the most actionable piece of the June 2026 data for public health planners. Misinformation-adjacent behavior correlates with provider access gaps as much as with platform choice. Interventions aimed purely at content moderation or AI output accuracy may miss a structural driver. The channel is a symptom; the access gap is the condition.

What the data don't yet tell us is whether AI health chatbot users are a distinct population from social media health users, or largely the same people using both. Disentangling that overlap matters for estimating the true denominator of myth-exposed adults, and for designing interventions that don't double-count reach. KFF's ongoing tracking series will likely be the fastest way to watch those numbers evolve.