New Poll Connects Social Media and Chatbots With Spread of Vaccine Misinformation

Americans who rely on social media and AI chatbots for health advice are significantly more likely to distrust vaccines, a new poll reveals. The survey, conducted by the nonprofit Annenberg Public Policy Center and shared exclusively with CNET, found that nearly 40% of adults who turn to social media for health information express vaccine hesitancy, compared with just 15% who rely on doctors. The link is even starker among users of generative AI tools: one in four chatbot users said they had encountered misinformation about vaccines, and half of that group subsequently doubted vaccine safety.

The findings arrive as health officials warn that the next pandemic threat could be as much about information warfare as disease. The poll surveyed 1,500 U.S. adults in February 2025, marking the first large-scale attempt to quantify how AI-driven platforms—particularly large language models like ChatGPT and Google Gemini—are reshaping public trust in immunization.

For years, social media has been a known vector for anti-vaccine propaganda, but the rise of conversational AI adds a new dimension. Chatbots can simulate empathy and authoritative tone, making false claims about vaccine ingredients, side effects, or efficacy more convincing. The Annenberg poll found that 22% of respondents had used a chatbot for health queries, and of those, 34% said the advice contradicted official CDC guidance.

Dr. Kathleen Hall Jamieson, director of the Annenberg Public Policy Center and lead author of the study, noted that the problem is compounded by the fact that chatbot responses often lack citations or disclaimers. Unlike a web search result, a chatbot’s narrative feels like an answer from a trusted expert. The poll also showed that exposure to vaccine misinformation on TikTok was 2.5 times higher among 18- to 29-year-olds than among older demographics.

Public health experts worry that AI-generated misinformation could reverse decades of progress in vaccination rates. The U.S. has already seen measles outbreaks in undervaccinated communities, and the World Health Organization lists vaccine hesitancy as one of the top ten global health threats. The Annenberg study suggests that if current trends continue, resistance to routine childhood vaccines could rise by an additional 15 percentage points within three years.

The implications for policy are clear. Lawmakers in Washington are debating the AI Accountability Act, which would require companies to label AI-generated health information and implement fact-checking safeguards. Meanwhile, platforms like OpenAI and Meta have pledged to improve content moderation, but critics argue that self-regulation has failed. The poll’s authors call for mandatory transparency: when a user asks a chatbot a health question, the model should cite sources and flag uncertainty.

Looking ahead, we can expect more research linking AI tools to health misinformation—and growing pressure on tech companies to redesign their products. The next big test will come with the respiratory season, when influenza and COVID-19 vaccines roll out. If the chatbots of 2025 cannot be trusted to provide accurate vaccine guidance, the public health cost may be irreversible.