Smartwatches that claim to read blood sugar through the skin: what the evidence shows

Smartwatches and smart rings sold across online marketplaces increasingly claim to measure your blood sugar through the skin, with no needle and no finger-prick. They do not work, and the US Food and Drug Administration has told consumers not to buy or use any of them: the agency has never authorised, cleared or approved a single smartwatch or smart ring to measure blood glucose on its own [s1]. The problem runs deeper than marketing — even research-grade non-invasive sensors, built with laser spectroscopy rather than a cheap optical sensor, still miss the true value by around a fifth, well outside the range you could safely dose insulin from [s2].

What these devices claim, and what they are

The pitch is a holy grail of diabetes technology: continuous glucose readings from a wrist or finger sensor, with nothing breaking the skin. It is important to separate two very different things. Legitimate continuous glucose monitors (CGMs) do pierce the skin — a tiny filament sits in the fluid just under it — and some display their readings on a smartwatch app; those are FDA-authorised devices [s1]. The products the FDA is warning about are the opposite: standalone watches and rings, sold under many brand names by dozens of companies, that claim to read glucose without any contact with blood or interstitial fluid [s1].

Non-invasive glucose sensing is a real scientific field — approaches use optical, spectroscopic and electromagnetic signals to infer glucose through the skin. But a 2026 review of the whole area is blunt about where it stands, listing the recurring obstacles: the variable relationship between glucose in blood and in other fluids, interference from ordinary physiological and environmental changes, calibration demands, motion artefacts, and, across the board, insufficient clinical validation [s3].

Why the FDA issued a warning

The risk is specific and serious. For someone with diabetes, an inaccurate glucose number can drive the wrong dose of insulin or of sulfonylureas — medications that lower blood sugar fast [s1]. Too much, acting on a falsely high reading, can send glucose dangerously low, and the FDA's communication spells out where that can lead: confusion, coma or death within hours of the error [s1]. That is why the agency's guidance is not "use with caution" but "do not buy or use," and why it applies to any such device regardless of manufacturer or brand [s1].

This is the rare consumer-tech story where the honest report is a harm report. A step-counter that is off by 15% is a minor annoyance; a glucose reading that is wrong by the same margin, trusted by someone deciding how much insulin to inject, is a medical hazard.

What the best non-invasive science actually achieves

To see how far even serious engineering is from the claim, consider a device that does the hard version properly. In a clinical test, a system using a quantum cascade laser to excite glucose molecules in the skin at infrared wavelengths — detecting the tiny heat signature they give off — was measured against blood glucose in 36 people [s2]. Its best algorithm reached a mean absolute relative difference of 20.7% in one session and 19.6% in another [s2]. The researchers frame that as roughly the accuracy of the earliest CGM systems the FDA cleared, and only for adjunctive use — meaning readings that must still be confirmed with a finger-prick before any treatment decision [s2].

Around 20% error is the current ceiling for a carefully built, laboratory-grade non-invasive approach. The consumer rings and watches the FDA warns about carry no such validation — the field's defining weakness, the 2026 review notes, is exactly this absence of clinical evidence [s3] — and none has been authorised to measure glucose at all [s1]. A device costing a fraction of a research instrument, with nothing published behind it, is not quietly matching that ceiling.

Why it matters

The market for consumer health sensors runs ahead of the evidence, and glucose is where that gap turns dangerous. It is the same pattern this site has traced in consumer glucose monitoring in people without diabetes, in general-wellness claims on wearables, and in direct-to-consumer genetic health reports: a device promises a clinical-grade answer that the underlying technology cannot yet deliver. The difference here is the stakes — the people most tempted by a needle-free glucose reading are precisely those who make dosing decisions from it. For what regulated glucose technology looks like, see our coverage of closed-loop insulin delivery, and for a parallel case of a consumer watch flagging a diagnosis, Apple Watch hypertension notifications.

What to watch

Whether any non-invasive glucose device earns genuine FDA authorisation — a specific, checkable milestone that none has crossed [s1] — and whether enforcement catches up with the sellers. Until a device clears that bar, the safe reading of a needle-free glucose watch is that its number means nothing, and that anyone managing diabetes should rely on an FDA-authorised meter or CGM chosen with their clinician [s1].

This article is informational and is not medical advice.

Sources

• [s1] "Do Not Use Smartwatches or Smart Rings to Measure Blood Glucose Levels: FDA Safety Communication." US Food and Drug Administration, issued 21 February 2024. https://www.fda.gov/medical-devices/safety-communications/do-not-use-smartwatches-or-smart-rings-measure-blood-glucose-levels-fda-safety-communication
• [s2] Kaluza M, Janik S, Lubinski T, et al. "Clinical validation of noninvasive blood glucose measurements by midinfrared spectroscopy." Communications Medicine, published online 15 November 2025. https://doi.org/10.1038/s43856-025-01241-7
• [s3] Imran A, Ramzan M, Hashmi L, et al. "Non-Invasive Technologies in Wearable Glucose Monitoring: A Structured Overview for the Future." Biosensors, published online 26 July 2026. https://doi.org/10.3390/bios16080407