What are the FDA or legal rules regarding AI rendering of cosmetics and supplements?

Quick Answer

The FDA and FTC regulate the claims your imagery makes about a cosmetic or supplement, not the rendering technique used to make it. An AI-generated image that implies a benefit the product cannot deliver, that suggests an ingredient the product does not contain, or that obscures legally required label text can be treated as deceptive advertising or misbranding under the FD&C Act and FTC Section 5. "Before and After" imagery should be avoided unless clearly labeled "Simulated."

This Is Not Legal Advice

This article is general information about how FDA and FTC rules apply to AI product imagery. It is not legal advice. Cosmetic and supplement claims are regulated areas with significant penalties. For decisions that affect your business, talk to counsel licensed in your jurisdiction.

The line that matters: cosmetic claim vs. drug claim

The FDA classifies a product by its intended use, and the imagery you publish is part of that intended use. A skincare product marketed as a cosmetic that "moisturizes" or "softens" is regulated as a cosmetic. The same product marketed with imagery implying it "treats acne," "regenerates cells," "reduces wrinkles," or "restores hair growth" can be reclassified as an unapproved new drug, which carries far heavier penalties. See the FDA's Is It a Cosmetic, a Drug, or Both? guidance.

For supplements, the analogous rule lives in DSHEA Section 403(r)(6). Structure/function claims ("supports immune health") are permitted with substantiation and the standard FDA disclaimer. Disease claims ("prevents flu," "lowers blood pressure") are not. Imagery that visualizes a disease outcome carries the same risk as text doing the same thing.

What the regulators look at in 2026

• MoCRA facility renewal, July 1, 2026: cosmetic manufacturers and processors must renew the facility registration they filed under the Modernization of Cosmetics Regulation Act by July 1, 2026, with biennial renewals after that. AI-rendered imagery does not change the registration obligation, but it can be evidence in a misbranding action.
• MoCRA cGMP rulemaking, by December 29, 2026: FDA is required to establish good manufacturing practice rules for cosmetics on this timeline; the proposed rule is expected late 2026.
• FTC Section 5, ongoing: deceptive acts in commerce, applied to imagery the same way it is to text.
• FTC Endorsement Guides (16 CFR Part 255), updated June 2023: any synthetic person used as if a real testimonial or customer needs a clear and conspicuous AI disclosure. The FTC's December 22, 2025 warning letters under the Consumer Review Rule signaled active enforcement on AI-generated review content.
• 21 CFR Part 111, supplements: cGMP requirements for dietary supplement manufacturing, including label accuracy. Imagery that misrepresents the labeled product can compound a label issue rather than fix it.

Critical guidelines for health and beauty brands

1. The "vignette" rule

FDA regulations allow vignettes (graphic scenes) on packaging and ads provided they are not misleading. The same standard governs AI imagery.

OK: a bottle of Vitamin C serum on a background of oranges (suggests the ingredient family).

Not OK: the same bottle next to a generated face with literally no pores or visible scars (implies a therapeutic benefit the product does not deliver).

2. Ingredient accuracy

If you use Nightjar to build a botanical-style background, the generated plants should match your actual ingredients. Aloe leaves next to a product that contains no aloe can be treated as a misleading representation of the product's composition. The same logic applies to any visible ingredient cue: lavender, rosemary, charcoal, vitamin droppers, or honey.

3. Label legibility

For supplements, the label text (Supplement Facts panel, allergen warnings, the DSHEA disclaimer where applicable) is legally required to be readable. Generic AI tends to blur or alter this text. Nightjar's Workflows (the product paths that govern how each Generation behaves, including Product Listing Image and Edit Images) are designed to preserve product content, including label text, text, and logos, so the legally required information stays readable in the final image.

A defensible workflow for cosmetics and supplements

Three habits keep AI imagery on the right side of the FDA and FTC lines:

• Anchor the image to a real product photograph. Generations that start from your actual product Asset (an image stored in Nightjar's Library) carry the real label and structure forward, so you are editing reality rather than inventing it.
• Lock model identity to one synthetic person. Nightjar uses a feature called Fashion Models: reusable AI people you select from a library of 80+ pre-built options or build yourself. A locked Fashion Model used across the catalog avoids the "every Generation is a new face" risk that increases right-of-publicity exposure and makes claim-substantiation arguments weaker.
• Save the setup as a Recipe. Nightjar's Recipes are saved Create-form setups (photography style, composition, model, background, output settings) that produce the same look across products. For a regulator, a stable Recipe is also an audit trail of how the image was made.

For ingredient and label fidelity work, the Edit tab's Edit Shortcuts (fast paths for common edits like Recolor, Reframe, and Change Format) keep the original product Asset intact and only modify the part you specify, rather than re-imagining the product from a prompt.

Summary

Use AI for the vibe (lighting, background, mood), not the evidence (results, body changes, ingredient claims you cannot substantiate). The FD&C Act and FTC rules apply to the image the same way they apply to the words around it.