When a Cosmetic Becomes a Drug: FDA's Intended Use Doctrine and Your Labeling Audit Framework

The line between “this moisturizer softens skin” and “this moisturizer repairs cellular damage to restore youthful structure” is approximately $500,000 in potential NDA filing costs and a complete reformulation of your go-to-market strategy.

That’s not hyperbole — it’s a description of what happens when FDA’s intended use doctrine reclassifies a personal care product as an unapproved new drug. And it happens more often than most regulatory teams expect, because the evidence FDA uses to establish intended use doesn’t come exclusively from your printed label. It comes from everywhere your marketing exists.

A brand can manage its primary display panel impeccably and still generate drug-claim evidence through a poorly worded Amazon bullet point, a paid influencer’s caption, or a press release that indexed on Google years before the current regulatory team joined the company. Most compliance audits don’t reach all of those channels. Enforcement does.

What 21 CFR 201.128 Actually Says — and What It Means in Practice

The core provision is brief. Under 21 CFR 201.128, a product’s intended use “may be shown by the circumstances surrounding the distribution of the article.” That phrase — circumstances surrounding distribution — is carrying most of the regulatory weight. It gives FDA authority to establish intended use from sources far beyond the label itself: advertising, sales force communications, trade show materials, paid influencer content, and customer service responses on a product’s website.

FDA defines a drug under Section 201(g)(1) of the FD&C Act as any article “intended to affect the structure or any function of the body.” A cosmetic, by contrast, is intended to cleanse, beautify, promote attractiveness, or alter appearance — without affecting structure or function. On paper, the boundary is clear. In marketing copy, it rarely stays that way.

The courts have consistently upheld FDA’s authority to use promotional materials as evidence of intended use. That creates a situation where a brand that files its product as a cosmetic, labels it as a cosmetic, and sells it as a cosmetic can still receive a warning letter for marketing an unapproved new drug — because someone in the e-commerce team wrote a claim that crossed the line.

The Modernization of Cosmetics Regulation Act, signed December 29, 2022, was the first major overhaul of federal cosmetic law in 87 years. MoCRA didn’t rewrite the intended use doctrine, but it significantly expanded FDA’s inspection authority and enforcement resources for cosmetic products. There are simply more reviewers examining more channels than there were five years ago.

The Four Claim Patterns That Trigger Drug Reclassification

Not all borderline language carries the same risk. After reviewing FDA enforcement letters and 483 observations across the cosmetics and personal care sector, four claim patterns appear consistently in products that get reclassified as unapproved new drugs.

Mechanism claims. Any language describing a physiological mechanism pushes directly into drug territory. “Stimulates collagen synthesis,” “activates cellular regeneration,” “modulates sebum production at the follicular level” — the word “activates” is a flag on its own. Paired with a biological target, it’s almost certainly a drug claim under the agency’s reading of 201(g)(1). The standard marketing instinct to sound scientific and credible is exactly the instinct that generates these.

Disease-state references. Using your product as a solution for “rosacea,” “eczema,” “psoriasis,” “hormonal acne,” or even “chronic dryness” invokes disease categories that require demonstrated safety and efficacy under 21 CFR Part 314 or, for certain OTC ingredients, an approved monograph. Most personal care categories don’t have a monograph to shelter under. Sunscreen and anti-dandruff do. A brightening serum doesn’t.

“Restore” and “repair” language. This is the single most common enforcement trigger we see in first-time audits at Aurora TIC. “Restores the skin barrier,” “repairs surface damage,” “rebuilds moisture levels” — all of these imply structural intervention. FDA’s reviewers interpret restoration as evidence of a prior impaired state, which implies a disease condition, which implies a drug claim. The distinction between “moisturizes” and “restores moisture barrier function” is small in most marketing departments’ minds. Under 21 CFR 201.128, it’s categorical.

Before-and-after imagery paired with clinical language. A photograph of visibly irritated skin transitioning to clear skin, placed adjacent to product usage instructions, functions as an implied disease-treatment claim even if the written copy stays cosmetic. FDA reviews the totality of the promotional unit — image, headline, subtext, and context — not the written claims in isolation. Your compliance review process should do the same.

Where FDA Finds the Evidence (And Where Your Audit Probably Isn’t Looking)

There are four channels that persistently underperform in standard promotional review programs.

Amazon product detail pages. The “From the Brand” section, bullet-point features, and customer Q&A responses are all promotional material. A manufacturer who responds to a customer question with “yes, this helps with hormonal acne” has added a drug claim to a public, indexed record. Amazon’s seller agreement doesn’t indemnify you from FDA enforcement. Neither does the fact that an account manager — not a regulatory professional — manages the listing.

Influencer and affiliate content. If your brand provided the product, scripted the talking points, or paid for placement, that content is attributable promotional material under FDA’s guidance on internet and social media promotion. FDA’s Office of Prescription Drug Promotion issues approximately 15 enforcement letters per year targeting prescription drug promoters specifically, and consumer safety officers routinely refer OTC and cosmetic violations from the same digital surveillance activity. The volume of influencer content being reviewed has grown considerably since 2022.

Archived press releases. Press releases from three or four product iterations ago frequently remain indexed by search engines and are publicly discoverable. A press release describing a clinical study that demonstrated your product “reduces inflammatory skin conditions” — written by a previous communications team, promoting a formulation you’ve since changed — is still evidence of intended use if a regulator finds it associated with your brand. We’ve seen this scenario in at least a dozen of the audits we’ve conducted.

Sales force training decks. For branded OTC products distributed through pharmacy or mass-retail channels, the sales representative’s leave-behind materials constitute labeling under 21 CFR 201.1. FDA inspectors request training decks during routine facility inspections. A claim on slide 14 of a 40-slide deck that hasn’t been reviewed since the product launched is fair game for a 483 observation.

Running an AI-Augmented Intended Use Audit

The core challenge with intended use audits isn’t knowing the rules. Every experienced regulatory professional knows 201.128. The challenge is volume. A mid-sized personal care brand might maintain 800 product listing variants across e-commerce platforms, 20 active influencer partnerships, a 300-page website, and a sales team using materials that predate the current QA director. Manual claim extraction across that footprint takes a compliance team six to eight weeks — a timeline that doesn’t account for the content being updated in real time.

AI-augmented audit approaches compress that timeline materially. At Aurora TIC, our claim-screening workflow applies NLP-based extraction to identify candidate drug claims across uploaded document sets — product labels, website crawls, marketing decks, Amazon listing exports, press release archives — and classifies them against a configurable claim taxonomy aligned to 21 CFR 201.128 and the relevant OTC monograph categories. A first-pass screen of 2,000 product pages that would occupy a compliance team for six weeks typically completes in under four hours.

That first pass is not the audit. It is the audit preparation. What the AI produces is a ranked list of flagged claims with citations to source document and applicable regulatory risk category. An experienced reviewer — someone who understands that “barrier function support” and “barrier restoration” carry different risk profiles — makes the final determination. The AI compresses the discovery window. The expert applies judgment.

Findings that surface consistently across brands undergoing this process for the first time:

• Amazon listing bullets authored by an e-commerce team with no regulatory review step in the workflow
• Influencer content that crossed from cosmetic into therapeutic language without triggering brand review, because the internal process only covered owned-channel content
• Archived press releases, still indexed, referencing claims the current label abandoned two reformulation cycles ago

The fix for all three of those is a documented claim management program with defined scope, channel-specific review authority, and periodic re-audits — not a single comprehensive review performed once at product launch.

Three Actions Before Your Next FDA Inspection

If you’re preparing for a pre-approval inspection, a facility inspection, or you simply want to tighten promotional review ahead of a product launch, three actions produce the most risk reduction per unit of effort.

1. Scope your claim inventory beyond the label. Build a systematic collection of all promotional materials, including third-party and archived digital channels. If your team can’t produce a complete, current claim inventory within 48 hours of a request, you cannot confidently defend an intended use challenge during an inspection.

2. Establish channel-specific review authority. Your SOP needs to specify who approves claims for Amazon listings, who approves influencer content briefs, and at what threshold a claim requires escalation to regulatory counsel. Most companies have a review process for traditional advertising and nothing documented for digital channels. That gap is precisely where exposure accumulates.

3. Build a claim taxonomy with documented escalation thresholds. Not every borderline phrase requires an attorney review. A three-tier taxonomy — cosmetic-safe, monitor, requires regulatory sign-off — gives your marketing team practical guidance and produces an auditable decision record. That record matters when a 483 observation questions your intended use controls.

Regulatory compliance consulting services that specialize in this area can accelerate all three, particularly if you’re launching into a new distribution channel or entering a product category where your team lacks deep precedent.

The intended use doctrine has been federal law since the original FD&C Act. What has changed is the number of channels it applies to and the AI-assisted monitoring resources that FDA and plaintiff’s attorneys are now using to surface violations systematically. Getting ahead of that gap costs a fraction of what responding to it does.

Written by Sam Sammane, Founder & CEO, Aurora TIC | Founder, Qalitex Group. Learn more about our team

Reserve early access to our AI audit tools Contact us

Related from our network

• ISO 17025-Accredited Label Claim Verification Testing — Qalitex Laboratories provides analytical testing to confirm that finished product formulations support the claims on your label, including OTC actives and cosmetic ingredients.
• Health Canada NHP and Cosmetic Labeling Compliance for Canadian Market Entry — Androxa supports regulatory submissions and testing requirements for natural health products and cosmetics entering the Canadian market under Health Canada’s framework.