EPA Intensifies Fluoride Safety Review Amid State Bans and AI’s Growing Influence on Health Information
The Environmental Protection Agency (EPA) is accelerating its review of fluoride safety, responding to increasing concerns – some fueled by misleading claims – and a growing movement to ban community water fluoridation. Simultaneously, the agency is grappling with the implications of artificial intelligence (AI) in disseminating health information, including the potential for both accurate guidance and the spread of misinformation.
Fluoride Safety Under Scrutiny
The EPA has released a preliminary assessment plan and literature survey as the first phase of its expedited review of fluoride safety. This action comes as several states, including Florida and Utah, have already banned community water fluoridation, with similar legislation proposed in at least 19 others. The review, initially scheduled for 2030, is now a priority for some within the Make America Healthy Again (MAHA) movement.
Despite the ongoing debate, water fluoridation remains a proven public health intervention, reducing tooth decay by more than 25% in both children and adults. However, concerns have been amplified by officials like HHS Secretary Robert F. Kennedy Jr., who have raised questions about fluoride’s safety without sufficient scientific backing. A 2024 National Toxicology Program report suggested a possible association between fluoride and lower IQ in children, but the studies analyzed were conducted outside the U.S. At fluoride levels more than twice the American standard.
The shift away from community water fluoridation is too impacting alternatives. The FDA has begun restricting some fluoride supplements, which opponents of fluoridation have previously promoted. Dental professionals are also reporting increased reluctance among parents and providers to use these supplements.
What to Watch For
The EPA’s ongoing review, regardless of its findings, risks eroding public trust in a long-standing public health practice. As alternatives face regulatory challenges, public confusion about fluoride’s safety may persist. Recent reports indicate a rise in emergency room visits for preventable tooth problems, a trend that could worsen as these narratives and policies gain traction.
AI and the Future of Health Information
The rise of AI tools like ChatGPT and Claude is transforming how people access health information. However, this shift also presents challenges, including the deliberate manipulation of AI responses through a practice called “generative engine optimization” (GEO). GEO involves structuring content to increase its likelihood of being surfaced by AI systems.
While GEO can facilitate disseminate accurate information, it also raises concerns about the potential for spreading false health claims. A recent study published in The Lancet Digital Health found that AI models were more likely to accept false medical recommendations presented in formal clinical language – such as hospital discharge notes – compared to information from sources like Reddit. Specifically, AI models accepted false recommendations in discharge notes 47% of the time, versus only 9% from Reddit posts.
Safeguards and Concerns
Researchers suggest that the authoritative tone of clinical language may lead AI models to apply less scrutiny to the information. This highlights the demand for context-aware safeguards, particularly in systems generating discharge recommendations or after-visit summaries. The EPA is also addressing AI integration within its own operations, developing an AI strategy to identify and remove barriers to responsible AI use, as outlined in its AI Strategy Plan.
FTC and AI Regulation
The Federal Trade Commission (FTC) is currently evaluating its authority to preempt state laws related to AI, following a presidential executive order. The administration argues that some state laws may require AI models to produce outputs deemed misleading, potentially stifling innovation. However, a bipartisan group of state attorneys general contends that many of these laws are consumer protection measures designed to combat deceptive practices like deepfakes and AI-generated scams. The FTC’s legal authority to preempt state laws is limited, suggesting a targeted approach to AI regulation focused on false advertising and deceptive practices.
X’s Experiment with AI-Assisted Fact-Checking
Social media platform X is testing a new feature that uses generative AI to propose fact-checks for Community Notes, which are then reviewed and edited by human contributors. AI-generated notes currently account for approximately 17% of Community Notes published on the platform. While X officials claim this collaboration improves speed and accuracy, some research suggests that AI may be replacing, rather than complementing, human fact-checking efforts.
What to Watch For
The effectiveness of AI-assisted Community Notes remains to be seen. Ongoing monitoring will be crucial to determine whether this approach improves the reliability of fact-checking on the platform and whether human involvement will continue to decline.