Is Your Health App a Medical Device? After the FDA's January Guidance, Your Marketing Page Decides

On 14 July 2025 the FDA issued WHOOP a warning letter over its Blood Pressure Insights feature. WHOOP argued publicly that the agency was overstepping. In June 2026 the FDA closed the letter, after WHOOP modified the product and its labelling [5] [6].

Five months before that closure, on 6 January 2026, the FDA published a revised General Wellness guidance whose Illustrative Example 7 is a wrist-worn wearable outputting blood pressure from a photoplethysmogram, described as meeting both factors for a low risk general wellness product, provided the values are validated [1].

Nothing about the sensor changed in between. What changed was where the boundary runs, and the boundary now runs almost entirely through language.

What the guidance actually did

General Wellness: Policy for Low Risk Devices was issued as final guidance on 6 January 2026, superseding the version issued on 27 September 2019 [1].

Commissioner Makary announced it at CES alongside the revised Clinical Decision Support guidance, framed as the agency adapting to AI and wearable technology. Contemporaneous law firm analyses of both documents are useful company to the primary text [3] [4].

The core framework is unchanged. CDRH “defines general wellness products as products that meet the following two factors: (1) are intended for only general wellness use, as defined in this guidance, and (2) present a low risk to the safety of users and other persons” [1]. For products meeting both, CDRH “does not intend to examine” them for device status or for compliance with registration and listing, premarket notification, labeling, quality management system or Medical Device Reporting requirements [1].

What is new is a passage on sensing. Quoted in full, because the conditions are the whole thing:

FDA may consider certain products that use non-invasive sensing (e.g. optical sensing) to estimate, infer, or output physiologic parameters (e.g. blood pressure, oxygen saturation, blood glucose, heart rate variability) to be general wellness products when such outputs are intended solely for wellness uses, and provided they:

• are non-invasive and not-implanted;
• do not involve an intervention or technology that may pose a risk to the safety of users or other persons if specific regulatory controls are not applied;
• are not intended for the diagnosis, cure, mitigation, prevention, or treatment of a disease or condition;
• are not intended to substitute for an FDA-authorized, cleared, or approved device;
• do not include claims, functionality, or outputs that prompt or guide specific clinical action or medical management; and
• do not include values that mimic those used clinically unless validated (e.g. manufacturer testing, peer-reviewed clinical literature) to reflect those values.

Then, immediately after, the permission that most coverage of this guidance skipped:

Products that meet the aforementioned criteria may display values, ranges, trends, baselines, or longitudinal summaries, and may contextualize these outputs in relation to sleep, activity, stress, recovery, or similar wellness domains.

And the boundary on the other side:

Products are not general wellness products when they are intended to measure, estimate, or report physiologic values for medical or clinical purposes, including screening, diagnosis, monitoring, alerting, or management of a disease or condition.

The two examples that draw the line

The guidance works through illustrative examples, and two of them are the ones this audience needs.

Example 7, on blood pressure. A wrist-worn product assessing activity and recovery, outputting hours slept, sleep quality, pulse rate and blood pressure, with sleep from an accelerometer and pulse and blood pressure from a photoplethysmogram. The FDA’s reading:

The claim relates to general wellness and does not refer to a specific disease or medical condition, and thus is a general wellness claim… Therefore, this product meets both factors for a low risk general wellness product, provided the product has validated values for blood pressure.

With a note attached: “if the claims made about any of the product’s functionality implied the product’s use in a medical or clinical context, the product would not be a low risk general wellness product” [1].

That is a wrist-worn PPG blood pressure feature, described by the FDA as acceptable under enforcement discretion, subject to validation. It is difficult to read that alongside the warning letter WHOOP received in July 2025 without concluding the ground moved.

Example 8, on glucose, is the one that shows the two factors are independent. A wearable estimating blood glucose for monitoring nutritional impact, measured via minimally invasive microneedle technology, explicitly contraindicated for use with diabetics and pre-diabetics, marketed for understanding insulin response to foods. The FDA’s reading splits:

This claim relates to a general understanding of a user’s health and is specifically contraindicated for use with a specific disease or conditions. Thus, it is a general wellness claim provided the product has validated values for blood glucose. However, the product penetrates the stratum corneum and is therefore not a low risk general wellness product.

So the claim passes and the risk factor fails, on the single fact that the sensor breaks the skin. Both factors have to hold. No amount of careful wording rescues a microneedle.

The CDS guidance probably does not apply to you

The two guidances shipped together and are constantly discussed together, which has produced a widespread misreading worth correcting. It is also worth getting the date right: the Clinical Decision Support Software guidance was reissued on 29 January 2026, superseding a version issued on 6 January 2026 [2]. The FDA revised it twice inside one month.

Non-Device CDS turns on the four criteria in section 520(o)(1)(E), and two of them are written around a healthcare professional [2]:

1. Not intended to acquire, process or analyze a medical image, or a signal from an in vitro diagnostic device, or a pattern or signal from a signal acquisition system
2. Intended for displaying, analyzing or printing medical information about a patient
3. Intended for supporting or providing recommendations to an HCP about prevention, diagnosis or treatment of a disease or condition
4. Intended for enabling an HCP to independently review the basis for those recommendations, so that it is not the intent that the HCP rely primarily on them

The abbreviation HCP appears 133 times in the document. If your product speaks to a consumer about their own sleep, activity or recovery, criteria 3 and 4 do not describe it, and the General Wellness policy is your document instead. A good deal of compliance content published this year applies these four criteria to consumer apps that are not within their scope.

For teams that do sell to clinicians, the substantive change is about single outputs. Previously, software presenting one recommendation rather than a set of options failed Criterion 3. The 2026 guidance states:

if only one option is clinically appropriate and the software function otherwise meets all criteria under section 520(o)(1)(E), FDA intends to exercise enforcement discretion (meaning that FDA does not intend to enforce requirements under the FD&C Act) for such functions.

The worked example is a function predicting risk of future cardiovascular events for an HCP to consider, based on weight, smoking status, blood pressure and a BNP test result. The sub-example that loses the discretion is the same function with variant genomic data added as an input “that does not have established relevance to the diagnostic recommendation” [2].

The claims are the product, regulatorily speaking

Intended use is established from labeling, instructions for use and promotional materials. That is not new, and it is the most under-appreciated fact in this area, because it means the regulatory status of your product is determined by artifacts your engineering team does not own: the landing page, the app store listing, the onboarding copy, the push notifications, the sales deck.

The same computation, with the same error characteristics, sits on either side of the line depending on the sentence wrapped around it.

None of those rewrites weakens a product. Several of them describe what the technology does more accurately than the version they replace, which is the recurring theme across everything we have written about health claims: the defensible wording and the honest wording are usually the same wording.

Less FDA does not mean less exposure

The tempting reading of January is that the regulatory burden on consumer health products went down. The burden did go down. What did not change is that the claims still have to be true, and the machinery for testing that is now less predictable.

Enforcement discretion means you do not need clearance. It also means you do not have a clearance to point at when someone challenges a claim. A 510(k) is expensive and slow, and at the end of it you hold a defensible, government-reviewed statement about what your product does. Enforcement discretion gives you freedom to make claims and no safe harbour for any of them. That is a structural observation, not a prediction: whatever the volume of challenges turns out to be, none of them can be answered by pointing at a clearance you were never required to obtain.

What occupies that space is the FTC, which requires competent and reliable scientific evidence for advertising claims, acts retrospectively, and concentrates on the loudest claims in a category. And it is private litigation under consumer protection law.

One live example, and it is an example rather than a trend. On 20 August 2026 a proposed class action was filed in the Northern District of California against Oura, alleging its marketing overstated how accurately its rings identify sleep stages, singling out a claim of 95% sleep staging accuracy against a clinical sleep lab. Oura disputes the allegations and notes the ring is not a medical device [7]. Nothing has been decided, and a complaint is an allegation.

The practical consequence is that the discipline the FDA used to impose externally now has to be imposed internally, and against a different standard. Not “would this survive review” but “could we substantiate this to a sceptical expert, in writing, today.”

What to do about it

Write down your intended use, in one paragraph, and circulate it. Every claim on every surface either supports it or contradicts it. Teams that cannot produce this paragraph on request generally have several intended uses running in parallel across marketing, product and sales.

Audit the surfaces, not the policy. App store description, landing page, onboarding, notifications, emails, sales collateral, support macros. The policy document is not where the regulatory risk lives.

Decide invasiveness early. It is the one factor claims cannot fix, and it constrains sensing strategy years before anyone writes a marketing page.

Substantiate before you publish, and keep the file. For every quantitative claim, record the task, the metric, the population and the reference. That is the file the FTC asks for and the file a plaintiff’s expert attacks. We set out what defensible accuracy wording looks like in where Oura’s 95% claim comes from.

Separate the US question from the EU one. Nothing here touches the AI Act, which runs on a different logic entirely, and where the same health score can be out of scope in a consumer app and inside Annex III when it informs insurance pricing. That is covered in the EU AI Act’s real dates.

The short version

On 6 January 2026 the FDA widened the General Wellness policy to cover non-invasive sensors that estimate physiological parameters including blood pressure and glucose, subject to conditions that are almost all about claims: no disease references, no diagnostic thresholds, no substitution positioning, no clinical alarms, no clinically-mimicking values without validation. Minimally invasive sensing is excluded on technology grounds regardless of claims. The Clinical Decision Support guidance, revised the same day, applies to professional-facing software and probably not to your consumer app at all.

The net effect is that the line between a wellness product and a regulated device now runs through your marketing page, and that a route which requires no clearance also leaves you with no clearance to cite when a claim is questioned. Whether challenges become more common is not something we can tell you. Where they land, the FTC and the courts, is.