Robert F. Kennedy Jr., the HHS secretary and longtime vaccine skeptic, outlined a sweeping plan to unify America's fragmented health data systems at a Maha Institute event on Monday. The vision: connect doctor visits, Medicaid records, Medicare claims, FDA surveillance data, electronic health records, pharmacy data, and immunization registries into a single AI-searchable infrastructure. The stated purpose is investigating chronic disease — but the operational focus, repeatedly and explicitly, is vaccines. The venue tells the story. The Maha Institute is led by Mark Gorton, who less than two weeks earlier declared himself 'the boldest anti-vaxxer in the anti-vax conference,' called vaccination success 'a complete fabrication' and 'massive fraud,' and claimed polio was caused by pesticides. The institute was co-founded by Tony Lyons, whose publishing house channeled $4m to Kennedy through Maha foundations funded by political contributions and health-related corporations, per the New York Times. Kennedy thanked Gorton, his 'longtime friend,' for his 'incredible support.' The data architecture Kennedy described is genuinely expansive. Medicaid covers 'hundreds of millions of lives, including tens of millions of children' tracked for 10-20 years. Medicare provides longitudinal data on older Americans. The FDA's Biologics Effectiveness and Safety (Best) System links claims data and electronic health records for 'active safety surveillance.' Commercial platforms like HealthVerity already link insurance claims, prescriptions, lab results, and EHR data. State-level datasets reported to the CDC, Kennedy complained, remain 'siloed' and 'unusable.' His goal: make it all 'research ready' and grant access to 'hundreds of external researchers.' The AI component is presented as transformative. Kennedy claimed studies that 'used to take years' can now be done 'literally in seconds.' He cited a meeting with OpenAI co-founder Sam Altman, who reportedly told Kennedy that HHS uses generative AI — Kennedy used the term 'superintelligence' — more than any other federal agency. This is the same OpenAI whose agents were recently revealed to have hacked US government websites and the Australian Medicare website. The specific research agenda Kennedy enumerated is almost entirely vaccine-focused: comparing vaccinated and unvaccinated populations using the CDC's Vaccine Safety Datalink; studying aluminum adjuvants; hepatitis B vaccine timing in infancy; live versus attenuated vaccines; influenza, Covid, and HPV vaccines in children; vaccination during pregnancy and childhood outcomes; maternal influenza vaccines and autism. An autism registry has been created 'despite widespread protest.' Kennedy returned repeatedly to a decades-old grievance about independent researchers being blocked from VSD data — referring to Mark and David Geier, whose medical license was suspended by Maryland's board of physicians, with David Geier charged with practicing medicine without a license. The structural concern is not data integration itself — better health data infrastructure is a legitimate policy goal. The concern is who controls the questions, who gets access, and what institutional safeguards survive. Kennedy explicitly wants to bypass what he calls 'gatekeepers' — the institutional review boards and confidentiality protections that exist to prevent exactly the kind of motivated reasoning his closest allies practice openly. When Gorton says outside organizations can 'fill the gaps' and 'bring political pressure,' the mechanism is clear: build the pipeline inside government, then let ideologically aligned outside groups run predetermined queries against the most sensitive health data in existence. The 20-year risk is not speculative. A health data system designed to confirm the priors of anti-vaccine activists, blessed by federal authority and turbo-charged by AI pattern-matching on hundreds of millions of records, would produce an endless stream of spurious correlations presented as causal findings. The damage to public health infrastructure — vaccination rates, institutional trust, evidence-based policy — could be generational. The data is real; the analytical framework being imposed on it is not.
