Two Regulators, Two Temperaments
In January, the Food and Drug Administration finalised guidance exempting low-risk wellness functions, including physiological trend analyses, from medical device oversight. The decision bears directly on the prevention mandate of public health systems. The rule is in force in the United States.
The Medicines and Healthcare products Regulatory Agency follows the opposite temperament. Its AI Airlock pilot remains in a second phase devoted to risk classification and post-market surveillance.
A dedicated "AI as a medical device" framework is expected for the United Kingdom this year, but is still awaiting publication. Britain proceeds cautiously. Washington moves at the speed of consumer electronics.
The US regulator has described its pace as aligned with the speed of consumer electronics. The British regulator prioritises risk classification before market access. Two philosophies, one product that crosses both.
The Product That Crosses Two Jurisdictions
Google's Health Guardian suite brings insulin-resistance trends, blood pressure monitoring and breathing analysis to the Pixel Watch and Fitbit. Pressure trends are derived from WavesFM, a foundation model trained on billions of minutes of cardiac and movement data.
According to the reconstruction by Distilled Post[1], the base model draws on patterns learned from more than one trillion minutes of data collected from five million users. The company frames every result as an insight, avoiding the word diagnosis. That language keeps the product at the margins of device law.
The device reaches the wearer's wrist next spring, according to the source. It delivers an autonomously generated message, without a clinical signature and without a GP referral. The watch has already set the priority.
Commercial convergence outpaces regulatory divergence. A single product, two legal regimes, an identical operational outcome for the patient. This asymmetry defines the governance problem.
The Mechanism: Why Language Matters
The choice of words determines the legal regime. "Insight" places the product in wellness; "diagnosis" would drag it into medical device law. The difference between the two words is worth the entire compliance architecture.
This mechanism produces a precise effect. The manufacturer retains commercial freedom, the regulator retains its formal jurisdiction, and the interpretive burden transfers to the public health system. Responsibility shifts towards the weakest link in the chain.
Healthcare organisations inherit that interpretive burden. The clinician receives data produced elsewhere, calibrated on a different population, and lacking the clinical context that would justify its use. The governance question arises precisely here.
The Governance Signal
The governance signal: a trend alert behaves like a diagnosis in the mind of the person who receives it, while remaining outside the clinical perimeter. Some patients will book an appointment, describe the data and expect the clinician to interpret information that the practice knows nothing about.
The burden is added to an already stretched primary care system. It arrives without a funding line, without a defined clinical pathway, and with an evidence base limited to validation studies conducted by a single company.
Volume matters. Five million users feeding a foundation model represent a scale that exceeds the response capacity of any primary care network. Every fraction that converts into an appointment multiplies the pressure.
The Quiet Erosion of the Prevention Mandate
The timing sharpens the argument. Yvette Cooper took on the health brief with the intention of relaunching a prevention agenda that, according to the King's Fund's assessment this summer, had stalled after the Tobacco and Vapes Act.
The 10-Year Health Plan commits the NHS to a shift from illness to prevention. A public prevention mandate presupposes that the health system defines what counts. The wristwatch has already decided what counts, and the system is still waiting to be consulted.
The timeline makes the tension visible. A public prevention agenda advances slowly, and a commercial product advances at the speed of the consumer market. The gap between the two paces defines the erosion described by the source.
The Alternative Reading
An opposing reading exists, supported by those who see preventive value in consumer data. An early alert can nudge the wearer towards diet, sleep and movement, objectives that the 10-Year Health Plan openly shares.
This reading holds on one condition: the existence of a clinical pathway that picks up the signal and translates it into action. In the absence of that pathway, the preventive benefit remains theoretical and the administrative burden remains real.
Reconciling the two readings requires an institutional choice. The health system decides to integrate the data within a funded pathway, or leaves it outside the clinical perimeter with clear communication to patients.
Three Decisions for the Board
Healthcare organisations and their legal advisers face three concrete questions. Each requires a written answer before the consumer data flow reaches the practice.
- General Counsel / Chief Compliance Officer: which internal role, named and in writing, is accountable for interpreting data generated by consumer devices?
- Chief Risk Officer: which risk framework covers the influx of trend alerts that lack a clinical pathway and dedicated funding?
- Board Audit & Risk Committee: which disclosure describes the organisation's exposure to data received by the practice on which an expectation of action still rests?
A compliance posture calibrated to medical device law is oversized relative to consumer wellness, and undersized relative to the real clinical burden. The audit remains required; the perimeter has changed.
Regulatory Horizon
Regulatory horizon: in the United States, the FDA guidance is in force. In the United Kingdom, the MHRA's AI Airlock pilot continues in its second phase, and the "AI as a medical device" framework remains expected within the year.
The question of whether wellness trends qualify as a device has been answered in the United States. A second question has opened in the United Kingdom, pending the promised framework. Organisations that build structured governance now, with named accountability and audit trails, will hold an 18-to-24-month advantage when enforcement begins.
Next spring marks the moment when the data flow reaches the market, according to the source. Governance decisions made before that date are meaningfully more useful than those made after.
This article was produced by an AI editorial author under human supervision, in accordance with the transparency obligations of Regulation (EU) 2024/1689 (AI Act, Art. 50). Sources are linked in the text.
Article by ATLAS
Sources
- Distilled Post (distilledpost.com)
- FDA – General Wellness: Policy for Low Risk Devices (Final Guidance, gennaio 2026) (fda.gov)
- GOV.UK / MHRA – AI Airlock: the regulatory sandbox for AIaMD (gov.uk)