The nationally representative survey of 2,480 US adults, conducted in May, found that people who use AI tools for health information at least once a week were significantly more likely to accept debunked claims about vaccines than those who never use the technology.
Researchers said the correlation remained even after accounting for factors such as age, race, education and political affiliation.
The findings come as AI becomes an increasingly popular source of medical information. A separate KFF survey published in March found around one-third of US adults have used artificial intelligence to seek health advice.
OpenAI has also acknowledged the growing demand, saying in a January blog post: "Health is already one of the most common ways people use ChatGPT, with hundreds of millions of people asking health and wellness questions each week."
Among frequent AI users, 35 per cent said they believed it was "definitely or probably true" that the measles, mumps and rubella (MMR) vaccine causes autism in children. That compared with 29 per cent of occasional AI users and 20 per cent of people who never use AI for health information.
The survey also found that 29 per cent of regular AI users believed mRNA vaccines can change a person's DNA, despite extensive scientific evidence disproving the claim. Among people who never use AI for health advice, the figure fell to 20 per cent.
Meanwhile, 22 per cent of frequent AI users believed the measles vaccine is more dangerous than the disease itself, compared with 15 per cent of non-users.
Researchers also identified similar trends among people who regularly seek health information on social media. Thirty-seven per cent believed the disproven claim linking MMR vaccines to autism, compared with 16 per cent of those who do not use social media for medical advice.
KFF noted that it did not ask respondents which AI chatbot they used, meaning the research could not assess the performance of individual models. Different AI systems can produce varying levels of misinformation depending on their training data and safeguards.
The findings add to growing concerns among researchers and health experts about the role AI could play in spreading medical misinformation as more people rely on chatbots for healthcare advice.
