WHOOP’s FDA Warning: The SaMD Line for Cardiac AI Investors

The line between a wellness feature and a regulated medical device is surprisingly easy to cross in AI health tech, and for product teams, buyers, and regulatory folks, knowing where that line is dictates product roadmaps, market access, and user safety. It’s not an academic exercise. A recent case involving WHOOP’s Blood Pressure Insights (BPI) feature is the perfect example of this distinction, showing how a product can stumble into diagnostic territory while the regulatory rules themselves are still in flux.

The WHOOP Blood Pressure Insights Feature and FDA Scrutiny

In July 2025, WHOOP got a warning letter from the FDA about its Blood Pressure Insights (BPI) feature, which brought a core tension to the surface: how do you give users useful health data without making claims that get you reclassified as a medical device? WHOOP’s position was that BPI was a wellness tool, designed to help users see how their bodies react to daily life, sleep, stress, exercise, not to diagnose anything Arnold & Porter advisory on FDA warning letter. This is the standard playbook for most wearable tech companies trying to help people with data on their own physiology. The FDA, however, saw it differently. They decided the feature’s claims suggested a diagnostic purpose related to hypertension, which yanked it out of the general wellness product category. That warning letter was a clear signal that the agency viewed the BPI feature as an unapproved medical device. For any team building AI features around vitals like blood pressure, this was a wake-up call that the FDA is watching for any claims that could lead a user to self-diagnose or change their medication without talking to a doctor.

Working through the “General Wellness” Category

The FDA’s general wellness policy is supposed to create a safe harbor for products that push a healthy lifestyle without making medical claims. These products generally either encourage wellness (like physical activity) or help people understand their health without diagnosing or treating a disease. The bright line is the complete absence of any intended use for diagnosis, cure, mitigation, treatment, or prevention of disease. The FDA’s issue with WHOOP’s BPI feature was all about the interpretation of its claims. Even if your team intends for a feature to be purely for “insights,” the language or functionality can implicitly suggest a connection to managing a disease state, and that’s enough to get it reclassified. This is especially sensitive for conditions like hypertension, where accurate diagnosis and management are critical for preventing strokes and heart attacks. Product teams now have to carefully vet every word in the user interface, in marketing materials, and even in internal documents to ensure claims stay firmly on the wellness side of the line. You have to avoid any hint of diagnostic capability. The risk that users will interpret data in ways a developer never intended (a form of algorithmic drift, really) is a constant challenge.

The FDA’s Evolving Stance: A Policy Shift

To make things even more complicated, the FDA’s own policy on general wellness products changed just a few months after the WHOOP letter. In January 2026, the FDA issued new guidance that actually widened the general wellness category, allowing some blood pressure-adjacent wellness claims but under even stricter conditions about what you can say. The revised policy doubles down on avoiding any disease, diagnostic, or clinical management claims. FDA Commissioner Marty Makary described the shift as an attempt to promote AI innovation while keeping high-risk products under full regulatory oversight MedTech Dive article on FDA policy shift. This shows the agency gets that technology is moving fast and regulation needs to be flexible but also strong. For product developers, this shift creates some breathing room and some new headaches. The expanded category might allow for more innovation, but the fundamental rule is the same: any claim that veers into diagnosis or treatment triggers medical device regulations, which means you’ll need rigorous clinical validation, a real QMS (like ISO 13485), and the right regulatory pathway, whether it’s a 510(k) or a De Novo classification.

Clarity in Classification: The Hello Heart Example

Unlike the regulatory friction WHOOP ran into, some companies are just very direct about their device classifications from the start. Hello Heart is a good model for this. The Hello Heart Monitor, which is the physical blood pressure cuff, is FDA-cleared as a Class II medical device. That means the hardware itself went through all the required regulatory hoops to prove its safety and effectiveness for measuring blood pressure. Here’s the important part: the company explicitly states that the Hello Heart app and its AI functionalities are not FDA-cleared. This distinction is key. It means that the measurement comes from a regulated device, but the interpretations and insights the app’s AI generates are carefully presented to avoid diagnostic claims, keeping them outside the scope of medical device rules. This kind of plain language about what is and isn’t cleared by the FDA prevents misinterpretation by both users and regulators. For product teams, this strategy of clearly segmenting the regulatory status of different components can be a powerful way to de-risk a project, especially when you’re adding AI to existing hardware.

Implications for Product Teams and Buyers

The WHOOP case and the FDA’s policy response offer some critical lessons for product teams building AI-powered health features. First, claim substantiation and language precision are paramount. Every single word in marketing, user interfaces, and product guides must be checked to ensure it can’t be misread as a diagnostic or treatment claim. What a developer intends doesn’t matter if a regulator or user sees it differently. Second, understand the evolving regulatory field. The FDA’s willingness to adapt its guidance on the general wellness category shows that what was acceptable yesterday might be re-evaluated tomorrow. Keeping up with these changes, and with principles like GMLP (Good Machine Learning Practice), is just part of long-term compliance now. Third, consider a modular regulatory strategy. As Hello Heart shows, clearly separating regulated hardware from non-regulated software/AI components, and being transparent about it, can create a much clearer path to market with less regulatory friction. For buyers, understanding these distinctions is just as important. A 510(k) clearance for a device’s measurement capability doesn’t automatically cover every AI-driven insight its app offers. Your due diligence has to include understanding the specific claims being made and the regulatory status of each component. Finally, the whole incident shows how important strong internal processes for regulatory review really are. Companies developing health AI have to bake regulatory literacy into their product development cycles from the beginning, designing for compliance instead of treating it as a last-minute check. That’s the only way to avoid the pitfalls of crossing the wellness-to-diagnostic line. AI in health has a ton of promise, but it’s also tangled in regulatory complexities. The WHOOP warning letter is a potent reminder that even well-intentioned wellness features can trigger significant regulatory scrutiny if their claims, however subtle, imply diagnostic intent. As the FDA continues to refine its approach to AI, the winners will be the ones who master clarity, precision, and a deep understanding of these regulatory boundaries.