The National Advertising Division (NAD) considered a recent case involving prescription hearing aids. This is not your typical toothpaste or toilet paper case we see at NAD but shows the forum welcomes any and all national advertising reviews. Before we get to the promised meat of AI substantiation, there was an interesting jurisdiction question we had not seen before.
This industry has a Hearing Industries Association, and both challenger and advertiser are members. The association has its own Claim Substantiation Guidelines and its own dispute resolution system, aptly named the Hearing Aid Advertising Review System (HEARS). The advertiser said as members they have an obligation to take grievances through their own association process. The challenger argued that the HEARS process is not available. NAD said regardless of the availability of a competing self-regulatory forum that NAD will hear all such challenges that come its way. NAD will administratively close a case that is before a court or the subject of an enforcement consent order, but not where another industry-specific challenge path is available. One open question is what would happen if NAD issued a decision suggesting changes be made to advertising while the HEARS board later issues a decision finding the claims supported. One option for the advertiser might be to pay to reopen a case to present new evidence to NAD. It is not entirely clear if a different decision from HEARS would be evidence, but we think it should be given that this is deeply technical tough stuff. This is all hypothetical of course, and case reopenings are relatively rare. But as NAD wonks interested in the procedural as well as the substantive, we thought it was worth the detour.
Omega AI claimed its hearing aids were “better than the rest in noisy conditions” and offered better “speech intelligibility.” There are industry-accepted tests involving humans used to validate such a comparison. Instead, the advertiser used AI speech-to-text modeling using a large, well-known and widely studied sound library to substantiate the claim, arguing that AI-based speech recognition is widely accepted in the field and such testing provides a more controlled comparison, eliminating any subjectivity or human variability, as well as allowing for the collection of many more data points than a human test subject would tolerate. The challenger argued that this library and these methods are used for speech recognition in the hearing world – not in the hearing loss context – and that speech recognition is not the same as speech intelligibility, which is the clinical benefit of hearing aids. The challenger argued overall that these lab predictions had not been clinically validated to translate into real-world outcomes. The advertiser agreed to modify and add “predictive” before the claims, signaling that AI modeling was the basis of support. NAD appreciated this modification but did not believe it would make clear to consumers that there was no human testing or validation that the testing applied in real-world conditions. NAD ultimately found the nascent technology was not competent and reliable to support a broad claim that promised measurable outcomes for wearers. NAD did conclude that nothing in the decision prevents the advertiser from “using AI technology to support carefully tailored claims.” The decision overall suggested that NAD is open to accepting AI-based substantiation as long as there is that all-important fit between the evidence and the claim being made. We’ll be keeping our ears to the ground to see if we hear more from NAD, hopefully loud and clear with actionable guidance, in other near-future cases looking at other uses of AI to support claims.
