Issue 017 traced who is writing the rules for AI in health practice, and the answer was a private coalition moving at vendor speed. This issue is about the government, because the last two weeks only make sense once you notice that it moved twice, in opposite directions. HHS committed federal AI capacity to research acceleration. The states closed out a half year with 84 new AI laws on the books, the fastest-growing category being chatbot safety. And in the gap between those two postures, the largest consumer AI company released its health product to every adult in the United States, one day after a patient’s lawsuit asked a court to stop it. Neither level of government examines your patient. The question this issue works through is what the split posture means for the one actor whose job is the individual in front of them.
Signal 1
HHS Joined the Genesis Mission and Launched AI Science Challenges With the White House, Committing Federal AI Capacity to Discovery, Not the Consumer Advice Lane (July 22, 2026)
On July 22, 2026, the Department of Health and Human Services announced it will participate in the Genesis Mission, the federal AI scientific research initiative created by executive order in November 2025, and unveiled a partnership with the White House Office of Science and Technology Policy (OSTP), the office that advises the President on science and technology, on a series of National Science and Technology Challenges. The flagship challenge directs next-generation AI at identifying root causes of chronic disease, with additional challenges on pediatric cancer research and drug discovery and development. The same day, OSTP released a report titled Science: A New Golden Age, framing AI-driven acceleration of research as a national priority. Source: American Hospital Association news, July 22, 2026.
What this means for you
Read a federal announcement by what it funds. This is the agency that held more than 7,000 comments asking it to govern clinical AI and answered with priorities rather than rules. Those priorities are now funded, and the funding went to discovery: chronic disease, pediatric cancer, drug development. Worth watching, because that research will eventually move the evidence base you practice from. None of it governs the AI operating inside practices this quarter. The operative consequence is one of sequence. If any part of your own AI policy is sitting open pending federal guidance, close it yourself. That guidance is not on the calendar, and every question it would have settled is now yours to settle.
Signal 2
OpenAI Released ChatGPT Health to Every Adult in the United States, One Day After a Lawsuit Asked a Court to Stop It (July 23, 2026)
On July 23, 2026, OpenAI announced that ChatGPT Health is now available to all users in the United States aged 18 and up. With the user’s consent, the feature connects medical records and Apple Health data to the chatbot, and it is integrated directly into the main chat experience rather than kept in a separate space, so connected health information can inform nearly anything the user asks. The company says more than 300 million people use ChatGPT every week to ask health-related questions, and acknowledges in the same announcement that the system can still make mistakes and that users should discuss medical decisions with their healthcare provider. The release came one day after a Florida man filed suit in San Francisco County Superior Court alleging that ChatGPT’s advice delayed his care ahead of a pulmonary embolism, a complaint that accuses OpenAI of unauthorized practice of medicine and asks the court to block ChatGPT Health pending a safety evaluation. Those allegations are untested, and OpenAI responded that ChatGPT is not a doctor and is not a substitute for medical care. The product is not an FDA-authorized medical device. Sources: OpenAI announcement, July 23, 2026; CBS News, July 22, 2026.
What this means for you
Read the sequence, because it tells you what actually governs this product. A complaint asked a court to halt it on Wednesday. It shipped to every adult in the country on Thursday. Nothing in the consumer lane compels a pause, and the only thing standing between the model and your patient is a terms-of-service line advising them not to rely on it. Set that against Issue 016, where the FDA-cleared patient-facing device is constrained to approved outputs for a defined population. That is the reviewed version, and it is deliberately narrow. This is the unreviewed version, and it now carries 300 million weekly health conversations with medical records attached. For your practice this changes intake, not policy. The patient across from you has likely already been advised, and that advice belongs in the history you take. Ask what they were told, and record it beside anything else they tried before booking.
Signal 3
The Mid-Year Tally: States Enacted 84 New AI Laws in Six Months, With Chatbot Safety the Fastest-Growing Category, Enforced Mostly by Attorneys General (Reported July 21, 2026)
On July 21, 2026, the Transparency Coalition published its 2026 Mid-Year AI Legislation Report: 84 new AI laws enacted in 27 states in the first half of the year, spanning chatbot safety, education and children’s digital lives, medical authorization and mental health, consumer rights, and frontier model oversight. Within that total, 14 chatbot safety measures passed in 13 states, and the movement is bipartisan, from Georgia and Idaho to Washington and New York. Hawaii’s SB 3001, requiring chatbot operators to disclose AI use and maintain protocols for responding to suicidal ideation, was enacted and took effect on July 14, 2026. Nebraska’s law bars a chatbot from indicating it is designed to provide professional mental or behavioral healthcare. Enforcement runs mostly through state attorneys general, with penalties up to 15,000 dollars per day in Rhode Island, while Oregon adds a private right of action for users who suffer harm. Source: Transparency Coalition 2026 Mid-Year AI Legislation Report, July 21, 2026.
What this means for you
This is where government actually reaches the consumer AI lane, and the shape of the reach is the point: state by state, after the fact, enforced mostly by attorneys general, with the strongest language aimed at minors and at anything representing itself as mental health care. The result is a patchwork in which the chatbot advising your patient can be lawful in one state and in violation across the border, and enforcement only engages once someone reports harm. Your exposure is therefore local, and your working knowledge should be too. Pull what your state enacted this year and file it with your scope-of-practice rules, because these statutes are starting to define what an AI system may claim about mental health care and what disclosure your patients are owed.
The Pattern
Government influence over AI in health is real, and it is running in two directions at once. Washington is accelerating: federal AI capacity just went to discovery, to chronic disease research, to a golden age framing. The states are refereeing: 84 laws in six months, attorney general enforcement, effective dates landing month by month. Between the two postures sits the vendor, who shipped a health product to every adult in the country the day after a court was asked to stop it, governed in the moment by nothing firmer than a disclaimer. Read the division of labor closely. The federal government is betting on what AI might find. The states are managing what AI already says. Neither one examines the patient in front of you, and the litigation that fills the gap only speaks after the harm. The practitioner is the one actor in this system whose entire job is the individual case, before the outcome, and the split posture of government is the strongest argument this Brief has yet recorded for governing your own lane instead of waiting for someone else to.
One Thing You Can Do This Week
Find your state in the mid-year tally and read what passed. It takes ten minutes: look up whether your state enacted an AI or chatbot law this year, read the one or two provisions that touch health practice or mental health claims, and write the effective dates into your calendar next to your license renewal dates. If your state passed nothing, record that too, because it means the only standard governing the chatbot advising your patients is the vendor’s own terms of service, and that fact belongs in how you ask patients about the advice they bring in.