Issue 014 asked who the author is when a machine drafts the record. This issue is about the calendar. On a single date, July 1, 2026, the Federal Trade Commission proposed its first policy statement on AI accuracy, three state laws governing AI in health practice took effect, and Medicare opened the first performance period of a model that pays for outcomes produced with digital and AI-enabled devices, some of them operating under FDA enforcement discretion rather than authorization. None of these are proposals anymore. The rules this Brief has been tracking for six months started arriving with effective dates. The question for a practitioner is no longer what is coming. It is which of the rules now in force reach the tools already sitting in your practice.
Signal 1
The FTC Proposed Its First Policy Statement on AI Accuracy, Treating Undisclosed Steering of a Model’s Outputs as Deception (July 1, 2026)
On July 1, 2026, the Federal Trade Commission released a proposed policy statement concerning what it calls the suppression of accuracy in artificial intelligence systems, and opened a public comment period that runs through July 31, 2026. The statement takes the position that AI companies have represented, explicitly and implicitly, that their systems are designed to produce the most accurate and faithful output possible within their constraints. Because consumers rely on those representations, and by the Commission’s account accept AI outputs without independent fact-checking more than 90 percent of the time, the FTC argues that steering a system’s outputs toward undisclosed objectives rather than toward what the user requested may be a deceptive act or practice under Section 5 of the FTC Act. The statement was issued pursuant to Executive Order 14365, signed December 11, 2025, which directed the FTC to clarify how Section 5 applies to AI models. Sources: Federal Trade Commission press release, July 1, 2026; proposed policy statement, FTC legal library, July 2026.
What this means for you
The federal consumer protection agency is now treating the fidelity of an AI system’s output as a legal representation, the same way it treats a health claim on a label. Your patients are among the consumers accepting AI answers without checking them, and any AI touchpoint your practice puts in front of a patient, an intake chatbot, an education tool, an automated follow-up, carries an implied accuracy claim that now has an enforcement theory attached to it. The comment window is open through the end of July, which means the professional bodies you belong to are writing their responses right now. What the final statement says about accuracy will shape what you can safely put a patient in front of.
Signal 2
Three State Laws Governing AI in Health Practice Took Effect on One Day: Tennessee on AI Therapist Claims, Indiana on AI in Claims Handling, Iowa on AI in Prior Authorization (July 1, 2026)
Three separate statutes crossed from passed to enforceable on July 1, 2026. Tennessee’s SB 1580 prohibits developers and deployers of AI systems, including chatbots, from advertising or representing that a system is qualified or capable of acting as a licensed mental or behavioral health professional, and treats violations as unfair or deceptive practices. Indiana’s HB 1271 works both sides of the claims table. It bars health insurers from using AI as the sole basis for downcoding a claim without a healthcare professional’s review of the medical record, and it bars providers from using AI to submit claims without review by the provider or a billing professional. Iowa’s House File 2635 permits utilization review organizations to use AI in prior authorization, but AI may not be the sole basis for a decision to deny, delay, or downgrade a request based on medical necessity. Sources: Holland & Knight, States Continue Efforts to Regulate AI in Healthcare, May 2026; National Law Review, state health AI legislation reviews, 2026.
What this means for you
Notice what Indiana did. Every prior statute in this space regulated the payer, the vendor, or the developer. Indiana’s review requirement reaches into the provider’s own billing workflow and makes a human look at what the AI prepared before it goes out the door under your name. That is the authorship principle from Issue 014 written into law, and it landed on the billing side first. If AI touches your coding or claims anywhere, the defensible posture in every state is the one Indiana now requires in statute. A named person reviews before submission, and your workflow documentation says so.
Signal 3
Medicare’s ACCESS Model Began Its First Performance Period, Paying for Outcomes From Digital and AI-Enabled Devices, Some Operating Under FDA Enforcement Discretion Rather Than Authorization (July 1, 2026)
On July 1, 2026, the Centers for Medicare and Medicaid Services opened the first performance period of the ACCESS model, a ten year voluntary model that pays Medicare Part B enrolled providers and suppliers outcome-aligned payments for managing hypertension, diabetes, chronic musculoskeletal pain, and depression with technology-enabled care. Alongside it runs the FDA’s TEMPO pilot, announced December 5, 2025, under which up to ten United States based manufacturers of digital health devices, including AI-enabled devices, may participate in ACCESS with products the FDA has not authorized for the intended use. During the pilot the FDA exercises enforcement discretion and does not enforce certain requirements against participating devices. The two programs are designed to work together, with the payment model generating the real-world evidence the FDA wants to observe. Sources: CMS ACCESS Technical Frequently Asked Questions, 2026; FDA press announcement, TEMPO pilot, December 5, 2025; Federal Register, December 8, 2025.
What this means for you
Read the two halves together. Medicare’s payment incentive for device-supported care switched on at the exact moment FDA oversight of some of those devices was deliberately lightened. Nothing about that is hidden, it is the design, and it moves the governance burden to the practice that supervises the treatment. If a device vendor approaches you citing ACCESS or Medicare outcome payments, the first question is now concrete. Is this device FDA-authorized for this intended use, or is it operating under TEMPO enforcement discretion. Those are two different risk postures, and the vendor knows which one applies. Get the answer in writing before the device touches a patient.
The Pattern
July 1 was not a news day. It was an effective date. The FTC now reads the accuracy of an AI system’s output as a legal representation. Three states now require a human to stand behind AI decisions in mental health representation, claims handling, and prior authorization. And Medicare began paying for AI-assisted outcomes under an FDA posture that is intentionally lighter than full authorization. Look at the direction of each move and a single design becomes visible. The rules are not slowing AI down. Restriction and permission arrived on the same day, and both point at the same place. Responsibility is being relocated onto the humans who deploy the tools. The regulators will let the AI in, and in some cases they will pay for it, on the condition that a named person reviews what it does and answers for it. That condition is no longer a professional norm taking shape. As of this month, in the first states and the first federal programs, it is the written floor.
One Thing You Can Do This Week
Take fifteen minutes and list every place AI currently touches your practice: anything patient-facing, anything in billing or coding, anything in documentation, any connected device. For each entry, write one line naming the person who reviews that output before it stands. If AI touches your claims, that line is now statute in Indiana and it is the defensible posture everywhere. If a line has no name on it, you have found the exact gap the July 1 rules were written to close, and you found it before an auditor, a payer, or a plaintiff’s attorney did.