Oura's 95% is a test result, not a sensor spec, and that is where the lawsuit lands

The number in dispute was never a hardware specification. "95% Sleep Staging Accuracy compared to clinical sleep lab" is the reported output of a comparison against something else ^[3], and the something else is polysomnography, which reads brain activity, eye movement and muscle tone through electrodes ^[4]. That construction is what makes the claim reachable. A vendor saying its ring estimates sleep stages is describing a feature. A vendor publishing an agreement percentage against a clinical reference has made a representation about a test, and tests have protocols, samples and scoring rules that a court can ask to see.

The sensor list explains why the percentage has to come from a model rather than a reading. The newest ring carries a photoplethysmogram using red and green LEDs for heart rate, blood oxygen and breathing, temperature sensors and an accelerometer ^[7]. None of those track brain activity, eye movement, electrical heart activity or muscle activity ^[8]. The complaint's phrasing for what fills the gap is "AI models... doing guesswork as to what may be happening, based on improper inputs" ^[9], and more bluntly, that "sleep happens in the brain, not on one's finger" ^[11].

The arithmetic is the part worth sitting with. Oura previously told customers 79% accuracy and more recently 95% ^[10], a climb of 16 points in the advertised figure ^[21]. The study the complaint leans on, covering 45 patients, put overall sleep stage classification accuracy at 53.18% and found REM overestimated by an average of 31.56 minutes ^[22]. Against the advertised figure that is a gap of 41.82 percentage points ^[20]. Neither percentage arrives with a stated protocol in the material available: Oura's answer is that its staging has compared favourably against polysomnography in multiple studies ^[23], which is a claim about a different set of comparisons rather than a rebuttal of that one.

Oura's fallback is already visible in the pleading, which alleges the company conceded the difference from a sleep study only in technical materials most customers will not read ^[15]. That is the standard shape of a California advertising case, and it is weaker here because the number was pushed through celebrity endorsements and influencer marketing ^[19] onto a device priced at $300 and up ^[16].

There is no regulator holding the other end of this. Wearables sit in a wellness category where accuracy can be supported by studies but the device is not treated by regulators as capable of medical diagnosis ^[18], so nobody certifies the percentage before it ships. The only venues that test it are advertising statutes, which is exactly what the seven counts do, including fraud by misrepresentation and California's Unfair Competition and False Advertising Laws ^[6].

The remedy is the expensive half. Damages are a number Oura can reserve against; a corrective advertising injunction requiring prominent disclaimers ^[6] rewrites the page that sells the ring, in the same year the company filed confidentially for a US IPO on an $11bn valuation ^[12]. Ryan Clarkson of the filing firm frames the harm as behavioural, saying people structure their days and interpret how they feel around the figures on screen ^[17], and years of users reporting that the ring called a bad night optimal ^[25] is the evidentiary base for that.