180 incidents on record · 2026 Headlights Incident reports by Ellie Harris · Melbourne
10 new this week Library last updated 30 August 2026
← The incident library
HD-INC-174
Digital health · United States · 2025 · AI health claims marketed as diagnostic without premarket authorisation

The FDA told a company its AI app was an unapproved medical device after it advertised a five minute gaze test to detect early signs of Alzheimer's disease

By Ellie Harris · Filed Inspection 24 to 26 March 2025; warning letter issued 21 August 2025

Alleged: SeniorLife Technologies, Inc. developed or deployed the AI system implicated in this incident. Details are drawn from public reports; parties are presumed innocent of any wrongdoing not established by an official finding.

The FDA told a company its AI app was an unapproved medical device after it advertised a five minute gaze test to detect early signs of Alzheimer's disease

What happened

It was reported that the United States Food and Drug Administration issued a warning letter on 21 August 2025 to a company whose mobile application used artificial intelligence to assess older adults. The letter quotes statements used to promote the product. They included that it conducts a five minute gaze based test on an iPad to detect early signs of Alzheimer’s, that it proactively identifies fall risk and recommends interventions, and that it was proven to reduce falls by 58 per cent in a two year Japanese study.

It was reported that the agency found the app was being promoted for uses that included screening and diagnosing disease, putting it outside the exercise equipment classification under which it had been registered. The letter states that the device was adulterated for want of premarket approval, that the website’s FDA Approved representation is false and its FDA Certified statement both false and misleading, and that no quality audit had been performed since the firm was established in 2022.

What an auditable version would have shown

The FDA letter says the claims appeared on the company’s website and app store listings. The sources reviewed do not show how many people used the product, how many were told they showed early signs of Alzheimer’s, or what happened afterwards. The letter quotes the claim that falls were reduced by 58 per cent in a two year Japanese study, but does not identify that study.

The company could have kept a record showing what evidence supported each health claim, what regulatory status the product held at the time and who approved the claim. For each assessment, it could also record what went into the model and what result was given to the carer.

Where the gap was

The FDA says the product was represented as FDA Approved even though it had not received premarket approval. It also says FDA Certified was false and misleading because the FDA had not certified the device.

A ConstraintGate is designed to check whether the organisation holds the authorisation behind a claim before it is published. A VerificationGate checks claims such as the 58 per cent figure against the evidence supporting them. A ConductRecord keeps a record of the claim, the evidence and what those checks found.

What governance should have looked like

The FDA found that the product had been registered under 21 CFR 890.5360 for Measuring Exercise Equipment, but that its promoted AI screening and diagnostic uses fell outside that classification. The letter also says the product had not received the required premarket approval.

Where an organisation makes health claims about an AI product, best practice would be to check those claims against the product’s actual regulatory status before publishing them. It should also hold the evidence behind any effectiveness claim and keep a record of who approved it.

Failure Pattern: the product was promoted as detecting early signs of Alzheimer’s and claimed a 58 per cent reduction in falls from a study that was not identified in the FDA letter. The FDA found that the product’s promoted uses fell outside the classification under which it had been registered.

Governance Principle: before making a health claim about an AI product, an organisation should be able to show the evidence behind the claim and that the product has the regulatory status the claim requires.

The reference implementation of ConstraintGate, VerificationGate and ConductRecord is open source. It lives at github.com/saffronandindia/headlights-oss, Apache 2.0 licensed and free to install. The repository is public now.

Sources

The mailing list

Fresh incident reports every week. One email to match.

We add new incidents to the library regularly, and send a single short email each week with what's new. The library stays free and open; this is just how you keep up with it.

No tracking. Unsubscribe in one click.

The record

An auditable system would have produced a signed, tamper-evident record the moment this happened: what the system did, the version that did it, the basis it acted on, and the action taken, and SeniorLife Technologies, Inc. could have produced it on demand.

This is the record the system as deployed did not produce in a signed, auditable form.

What this teaches
Capture what happened when it happens
What the system did, the version that did it, the basis it acted on, and the action taken, recorded at the moment, not reconstructed after.
Sign it, so no one has to trust the record-keeper
A tamper-evident entry. Edit it later and the signature breaks. The record does not ask for the benefit of the doubt.
Make it verifiable by anyone
A court, a regulator, a customer's lawyer can check the record themselves, without taking the company, or us, at our word.

Headlights summarises publicly reported AI incidents. All summaries are independently written, attributed to their original sources, and intended for research and educational purposes. Allegations are identified as such until established through official findings.

Last reviewed August 2026. This report is based on the sources listed above and reflects information available at the time of review; later developments may not be captured. Where a person is described as charged with or alleged to have done something, that allegation is unproven unless a conviction or a court or regulatory finding is stated. Headlights publishes journalism and commentary, not legal advice.

Want to write back?

Direct to my inbox.

ellie@useheadlights.com →