More Americans are asking AI chatbots for health advice. They're falling for the same false claims as social media users.

The numbers

One in three American adults now ask AI chatbots for advice about their physical or mental health, according to polling data released by KFF in March 2026. That puts AI roughly level with social media as a source for health information.

But here's what should catch your attention: people who use these AI chatbots regularly are believing false claims about vaccines at almost the same rate as people who get their health information from social media. A follow-up KFF poll from June 2026 found that 35% of adults who use AI chatbots weekly for health information believed vaccine myths. Among weekly social media users seeking health information, it was 37%.

The numbers are nearly identical.

The surprise

When AI chatbots first emerged, many expected them to be more reliable than social media. Unlike Twitter or Facebook — where anyone can post anything — AI systems are trained on medical literature and, in some cases, official medical databases. They're built by engineers and trained by specialists, not left to run wild like a social media feed.

The KFF data suggest that advantage doesn't exist, at least not yet. Both channels appear to spread health misinformation at the same pace.

There's another pattern worth noting. The same polling found that people without a regular doctor or nurse — no trusted healthcare provider — are more likely to believe vaccine myths, whether they get information from AI or social media. This overlap creates a particular vulnerability: people without access to steady medical care are more likely to turn to these digital sources, and those sources aren't reliably steering them right.

The trust problem

There's another gap between what people do and what they actually feel comfortable with. Only about one in three adults say they would trust an AI tool to access their personal medical records, according to KFF polling from January 2026. People are using these tools without full confidence in them — a pattern we've seen before with earlier health technologies like patient portals and at-home genetic tests. Adoption moves faster than trust does.

What comes next

For public health officials and doctors, the implication is straightforward: AI chatbots are no longer a niche thing. They're already a main source of health information for millions of Americans. That scale demands the same attention that regulators and health authorities gave to social media over the last decade.

The June data hint at something worth taking seriously. The problem may not be the technology itself — not specifically how AI chatbots work or how they're built. The correlation between misinformation belief and lack of a trusted doctor suggests a simpler issue: people who don't have regular access to a doctor they trust are seeking answers somewhere, and both AI and social media are filling that gap. When the real problem is a missing relationship with a healthcare provider, only changing the channel won't fix it.

That's worth holding onto, because it changes where public health efforts should focus. Instead of purely trying to clean up what AI chatbots say or improve their accuracy — important as that is — there may be real value in addressing the underlying gap: the people who don't have steady access to a doctor in the first place.

One question remains unanswered: Are AI chatbot users mostly different people from social media health users, or are they largely the same people using both? That matters. If they're the same people switching between platforms, the actual number of Americans at risk is smaller than the statistics might suggest at first glance. If they're different groups, the risk is broader. Future tracking polls will help clarify that picture.