The Ask Cass Guide · AI x Technology

The AI x Beauty Law Guide

AI in beauty ads: the disclosure laws you need to know.

Three states now require some ads to say when AI made or changed what you're looking at. None of them were written with beauty in mind, and beauty is where they'll get tested first: before-and-afters, product demos, flawless skin. Here's what each law actually covers, where the gaps are, and what to ask before your next campaign.

By Cassandra Campbell, Editor · Updated November 12, 2026 · This is news, not legal advice.

The short version

New York and California are about fake people. If the person demonstrating your serum or narrating your ad was generated by AI, the ad has to say so. Pennsylvania is about fake results. If AI makes the product look like it does something it doesn't, the ad needs a label whether or not a person is in it.

For beauty, Pennsylvania may matter most, because the industry's oldest problem is the image that oversells the result. AI just makes that image cheaper to produce.

The three laws

In effect since June 9, 2026

New York

Synthetic performer law (S.8420-A)

What it covers
Ads with an AI or computer-made person who isn't a real, recognizable performer. Audio-only ads and AI used just to translate a real performer are left out.
Who's responsible
Whoever makes the ad, and only if they actually know the performer is synthetic. Publishers that only run the ad are exempt.
Penalty
$1,000 for a first violation, $5,000 after that. No right for consumers to sue, but the attorney general's office was already getting consumer complaints by September.

Takes effect January 1, 2027

California

SB 1050

What it covers
Ads where an AI-generated person demonstrates the product, narrates or voices the message, or reacts to it. Includes audio-only ads. Someone in the background may not count.
Who's responsible
Whoever creates the ad and causes it to run, with no requirement that they knew. Platforms have to pull an ad once a court orders it.
Penalty
No set fine. It's enforced through the state's false advertising law, which means consumers who lost money can sue, along with prosecutors.
Follow this story →

Takes effect December 21, 2026

Pennsylvania

Act 99 of 2026 (SB 806)

What it covers
Ads where AI-generated or heavily AI-altered images, video or audio give a false impression of the product. Text-only ads, parody, satire, commentary and obvious fiction are exempt, and political ads aren't covered.
Who's responsible
Advertisers who knowingly, recklessly or willfully run a misleading AI ad. Only the attorney general or a district attorney can enforce it.
Penalty
Up to $3,000 plus legal costs. The law doesn't say whether that's per ad or per violation. No right for consumers to sue.
Follow this story →

Where beauty gets tricky

An AI-smoothed before-and-after

This is the clearest case for Pennsylvania. If AI evens out the “after” skin beyond what the product did, that's a false impression of the product. New York and California likely don't apply, because the person is real.

An AI avatar applying your serum

Covered in New York now and in California from January, since the performer is demonstrating the product. California suggests wording like “this performance features a synthetic performer.”

An AI voice reading the ad

California covers audio-only ads and synthetic narration. New York doesn't cover audio-only ads.

A real creator, edited by AI

This is the gray area. The New York and California laws are about performers who aren't recognizable as a real person, so a creator whose face was smoothed or reshaped may fall outside them. Pennsylvania could apply if the edit misrepresents the product. The bigger issue is consent: a creator in California has sued over an ad that used AI to lighten her skin, and the app blamed automatic ad enhancements. Lawyers expect this kind of edit to be where the first disputes land. Hawaii already points that way: since July, a person there can sue if a realistic AI imitation of them is used in an ad without their consent, for up to $25,000 per ad.

A filter on a sponsored post

No US law here names filters. The UK's ad regulator ruled in 2021 that filters which exaggerate what a beauty product does are misleading, and that the brand is responsible. Pennsylvania's law points the same way when the filter is AI.

Rules that already apply everywhere

The state laws are new, but the federal baseline isn't. Since 2023, the FTC's endorsement rules have treated virtual influencers like human ones: the paid relationship has to be disclosed, and a virtual endorser can't imply it used a product that only a person can use. The FTC's fake review rule bans AI-written reviews and testimonials, and lawyers covering it name beauty as one of the most exposed industries.

If you sell in Europe, the EU AI Act has required deepfakes to be clearly labeled since August 2, 2026. And in California, from January 1, the state's likeness law also covers AI replicas of a real person used in ads without consent.

One thing to watch: a December 2025 executive order told the Justice Department to challenge state AI laws the administration sees as burdensome, and in April the department moved to join a lawsuit against Colorado's AI law. So far none of the ad disclosure laws on this page has been named as a target, so plan on following them.

Before your next campaign

If you're a brand or agency

Can you say where AI touched every asset in this campaign, including tools that edit automatically?

Is any AI imagery close to the product result, like skin, hair, lashes or shade?

Do your creator contracts say whether AI edits are allowed, and who approves them?

If an ad runs nationally, are you writing to California's rules, the strictest of the three?

If you're a creator

Does your agreement say the brand can't alter your face, skin or voice, with AI or otherwise?

Does it limit where your content runs as a paid ad, and for how long?

If you use AI in your own sponsored content, would a viewer get the wrong idea about what the product does?

Why I'm watching this

I've worked on all sides of how beauty gets sold: in the makeup chair, inside a brand, and at the platforms that run the ads. The gap between the image and the product has always been there. What's new is that a machine can now widen it in seconds, at scale, without anyone deciding to.

These laws are a first attempt at closing that gap, and they're uneven. None of them has been tested in court yet. I'll update this page as they take effect and as the first cases come in.

Coming up

  • December 2, 2026 EU deadline for AI providers to mark generated content in a machine-readable way.
  • December 21, 2026 Pennsylvania's Act 99 takes effect.
  • January 1, 2027 California's SB 1050 and SB 1111 take effect.
  • Waiting The federal NO FAKES Act, which would create a national right over your face and voice, is stalled in the Senate.

Sources

Cooley on New York's law · Mayer Brown on SB 1050 · Vorys on SB 1050 lawsuits · The Next Web on Act 99 · Let's Data Science on Act 99 · BakerHostetler on the FTC endorsement rules · UK ASA on beauty filters · Silicon Canals on the EU AI Act · WHYY on Pennsylvania's law · Net Influencer on New York complaints · Frankfurt Kurnit on Hawaii's law · Jenner & Block on the DOJ and Colorado

Ask Cass summarizes public reporting and legal analysis. This is news, not legal advice. If you're planning a campaign, talk to a lawyer about your specific situation.