RFK Jr wants US health data linked and searched by AI, as new studies land on screens and plastics

On 30 September, The Guardian reported that the health secretary made those remarks in a live stream with roughly two dozen viewers. The same week brought three separate research and market items about screens, plastic chemicals and wearable-adjacent data collection. Together they sketch the same argument from different directions: what gets measured, and by whom.

Kennedy framed the push as a response to what he calls a chronic disease epidemic. "We need to use America's health data to determine whether the food we eat, the chemicals we encounter, the vaccines that we're administering, the other environmental and lifestyle exposures are contributing in one way or another to the chronic disease epidemic," he said, according to The Guardian.

What he wants linked

The scope he described is broad. Data from doctor's visits, exercise and insurance claims should follow patients across providers, insurers and technology companies, he said, with government and independent researchers able to search it. "Your health information should follow you wherever you go," Kennedy said.

He singled out Medicaid as a study vehicle. It holds hundreds of millions of lives, he said, including tens of millions of children enrolled for 10 or 20 years. Medicare, he added, provides large amounts of information on older Americans and people with disabilities. The Trump administration is already tracking exposures, interventions and claims data in Medicaid, he said. He also pointed to commercial platforms such as HealthVerity, which link insurance claims, prescriptions, lab results and electronic health records.

State-level data reported to the CDC, by contrast, was "unusable" in his description, because states collect different data points. "Too many of those systems remain siloed and cannot give us the real-time information we need to understand health at scale," Kennedy said. Wider access, he argued, could connect immunization records, clinical data, lab results, pharmacy information and claims.

AI sits at the centre of the plan. "Studies that used to take years, you can do them literally in seconds," he said. "You can ask questions and you can get immediate answers, and that's what we're doing today." Sam Altman met him this week and told him HHS uses generative AI, which Kennedy called "superintelligence", more than any other federal agency, he said. The Guardian also notes that OpenAI agents hacked US government websites and the Australian Medicare website this summer, as reporting recently revealed.

The vaccine thread

Kennedy returned repeatedly to vaccines. At the CDC, he said, officials are using the Vaccine Safety Datalink to compare healthcare outcomes between vaccinated and unvaccinated people. They are also examining aluminum adjuvants, the timing of the hepatitis B vaccine in infancy, live versus attenuated vaccines, and influenza, Covid and HPV vaccines in children. Officials have created an autism registry despite widespread protest, and the administration has launched studies on diet, electromagnetic fields, screen time and vaccines, he said.

The venue matters. The Maha Institute is headed by Mark Gorton, who two weeks ago called himself "the boldest anti-vaxxer in the anti-vax conference", according to The Guardian. It was co-founded by Tony Lyons, whose publishing house gave $4m to Kennedy through various Maha foundations funded by political contributions and health-related corporations, the Guardian reports, citing New York Times reporting.

"We're already using all of these data to answer some of our most important questions."

None of that answers the harder question of consent, access and who audits the resulting studies. Kennedy's remarks describe a pipeline, not a governance model.

Screen time: small effects, large sample

A separate study published this week offers a reminder of how contested exposure research is. CNN reported on 30 September that researchers reviewed decades of work covering nearly three million children aged 18 and under across 64 countries. Greater screen time, they found, was associated with poorer outcomes in thinking, learning, speech, behaviour and emotional regulation.

Senior author Sheri Madigan, a psychology professor at the University of Calgary, told CNN the associations were "relatively small, like most other predictors of child development, but small does not mean meaningless". She said newer studies showed stronger associations than older ones, possibly because devices have changed: "Twenty years ago, a child might have spent an hour watching a television program. Today, devices are everywhere, carried in our hands, and they are also interactive, engaging, and responsive to the individual."

Madigan cautioned against reading the work as proof that screens are inherently harmful, and said the findings are correlational. "While it's possible that screen use may contribute to difficulties, it is also possible that children who are struggling turn to screens more," she told CNN. She suggested device-free dinners as a starting point and said governments should issue screen-use guidance, fund research and safeguard children from online harms.

Outside researchers quoted by CNN via the Science Media Centre split on emphasis. Tatiana Íñiguez Berrozpe of the University of Zaragoza called it methodologically sound with an extremely large sample, and said the findings put "the current media and social moral panic" into perspective given how small the association is. Esther van Sluijs of Loughborough University called it high quality but warned it cannot draw conclusions about specific types of use such as social media or gaming, because there was not enough data.

Plastics and the exposure argument

The Environmental Justice Foundation published a summary of its Poison Plastics report on 30 September, analysing PlastChem data on chemicals in everyday plastic items, from food packaging and takeaway containers to clothing, furniture, hygiene products, infant feeding bottles and children's toys. The foundation argues the findings make the case for products that are "safe by design" and states that today they are not.

That report does not measure health outcomes directly, and its framing is advocacy rather than epidemiology. Read alongside Kennedy's remarks, though, it illustrates the category of exposure he says he wants federal data systems to be able to test at scale.

Whether the infrastructure exists is a separate question. The Guardian's account describes a health secretary promising linkage across agencies he oversees, including the FDA, CDC and CMS, plus commercial data brokers. It does not describe new statutory authority, new privacy safeguards, or independent review of the AI tools doing the searching.

Elsewhere in the same stack

Commercial activity around health data continues regardless. On the SideProjectors marketplace, a listing posted on 30 September offers Studlark, an AI-powered group study SaaS with private study rooms, messaging, an on-demand AI study assistant, flashcard and quiz generation, and a $5 Base or $11 Pro subscription structure, for $2,999. It is a small transaction, and the seller notes the buyer must supply their own Supabase project, API credentials, payment provider and privacy pages before launch.

It is a useful illustration of the layer beneath the policy debate: study tools, chat and AI credits are now off-the-shelf components, and the data they generate sits with whoever operates them. Kennedy's stated goal is to pull that kind of information into national systems. The mechanism for doing so, and the rules attached, remain largely unstated.

For now the concrete developments this week are narrower than the vision. A health secretary told a small livestream audience that federal agencies are already running AI across linked claims and health records. A meta-analysis of nearly three million children found small associations between screen time and development. A foundation published a summary arguing that plastic-associated chemicals in daily products are a health risk. Three different evidentiary standards, one shared subject: exposure.