Total:
$0.00
Here is a campaign an eyewear store ran last month, and it is probably close to one you have run yourself.
A generative model produced a woman in her thirties, high cheekbones, nobody in particular, wearing your bestselling tortoiseshell round frames. She appears as the second image on the product page. She appears again in a six-second Meta ad where she turns her head and smiles. And a synthetic voice reads the offer over a fifteen-second spot on a podcast network.
One face, one afternoon, one invoice. As of 1 January 2027, those three assets sit in three different legal positions in two different states, and the thing that separates them is not what the model looks like. It is whether the asset moves, and whether it makes a sound.
What California signed on 16 September
SB 1050 (Ashby) was approved by the Governor and chaptered on 16 September 2026 as Chapter 246, Statutes of 2026. It adds Article 10, commencing with section 17610, to the false advertising part of the Business and Professions Code. It is the second state synthetic performer disclosure law in the United States, after New York.
The operative sentence is one line:
It is unlawful for any person to create and cause to be published in an advertising medium an advertisement that prominently includes a synthetic performer without a clear and conspicuous disclosure that the advertisement includes a synthetic performer.
The bill carries no urgency clause, so it takes effect under the ordinary rule for statutes passed in a regular session: 1 January 2027. You will see commentary describing it as effective on signature. That is a reading of the signing date, not of the statute.
The required wording is prescribed, which is unusual and helpful. Subdivision (c) says the disclosure shall use wording substantially similar to "this performance features a synthetic performer" or "no human performer is depicted". You do not have to invent a phrase and hope.
The definition is wider than the duty, and that is the whole story
Read the definitions in the order the statute puts them, because the gap between two of them is where every practical question in this article lives.
A synthetic performer is defined at section 17610(a)(6) as:
a digital figure, voice, or representation created in whole or in part using generative artificial intelligence that creates the realistic impression of the audio, audiovisual, or visual performance of a human performer who is not recognizable as any identifiable natural person.
Note the word visual. A still image of a person who does not exist is squarely inside that definition.
Now read what the duty attaches to. An advertisement, at 17610(a)(1), is:
any audio, video, or audiovisual message, statement, audiovisual recording, digital communication, or other representation disseminated in any manner or by any means, including through online platforms, that is intended to induce, or that is reasonably expected to induce, the purchase of goods or services.
And an advertising medium, at 17610(a)(2)(A), is:
any broadcast station, cable operator, multichannel video programming distributor, online platform, streaming service, digital advertising network, publisher, or other person or entity that distributes, displays, transmits, or makes available an audio or audiovisual advertisement to consumers in this state.
So the class of things that can be a synthetic performer includes a still visual performance. The class of things that can be an advertisement does not. Both gates have to open for the duty to bite, and for a photograph the second one stays shut.
The practical consequence for an eyewear store is blunt. The synthetic model on your product page is not what this law is about. The same synthetic model in your video ad is. A single generated frame of her is outside; twenty-four of them a second is inside.
New York and California are mirror images
This would be a footnote if California were the only state in play. It is not. New York's synthetic performer law has been in force since 9 June 2026 — we covered what it means for eyewear product photography when it landed, along with Amazon's metadata tag and California's provenance rules.
The two statutes were written eight months apart, both by people looking at the same problem, and they draw the line in opposite places.
| New York (GBL § 396-b, in force 9 June 2026) | California (B&P § 17610, from 1 January 2027) | |
|---|---|---|
| Still image ad | Within the definition — "audiovisual and/or visual performance" | Outside — an "advertisement" must be audio, video or audiovisual |
| Audio-only ad | Expressly exempt | Expressly covered — "audio" leads both definitions |
| Video ad | Covered | Covered |
| Does the merchant need to know? | Yes — duty applies "where such person has actual knowledge" | No knowledge element in the section |
| Threshold | Any use | Must "prominently include" the performer |
| Wording | "Conspicuously disclose" — undefined | Prescribed phrases, and "clear and conspicuous" is defined |
| Who is bound | A person who "produces or creates" an advertisement | A person who "create[s] and cause[s] to be published" |
| Penalty | $1,000 first violation, $5,000 subsequent | Treated as a § 17500 violation; see below |
A Shopify store sells into both states from the same catalogue, with the same creative, on the same day. There is no version of this where you comply with one and ignore the other, and there is no single rule that satisfies both by accident. The honest summary is that if any asset in a campaign contains a generated person, the cheapest answer is to disclose on all of them — which is roughly what the two states, between them, have engineered.
"Create and cause to be published" is two verbs, not one
This is the sentence to take to whoever makes your video.
California binds a person who does both things: creates the advertisement and causes it to be published in an advertising medium. New York binds a person who produces or creates one — either will do.
Three common arrangements, and they do not land in the same place:
You generate the video yourself and upload it to Meta. You created it and you caused it to be published. Both verbs. This is the clear case and it is also, increasingly, the normal case, because generating a six-second product video no longer requires anyone but you.
Your agency generates it and you approve and run it. The agency created it. You caused it to be published. Whether either of you satisfies the conjunctive test on your own is a real question, and it is the kind of question that gets resolved by whoever has the better indemnity clause rather than by a court. Fix it in the contract at renewal, not afterwards.
You licence AI-generated stock footage and cut it into your ad. You did not create the performer. You did assemble and publish the advertisement. The argument that you are outside subdivision (b) exists; it is not one I would want to be making to the Attorney General while the alternative was a seven-word caption.
Note also who is not the duty-holder. Meta, YouTube, a podcast network and a streaming service are the advertising medium — they are the place the ad is published, not the person bound by subdivision (b). Their obligation, at subdivision (e), starts only once a court has found a specific advertisement unlawful, at which point they must remove it or cease further dissemination. The platform is the venue. You are the person.
The duty attaches because she is nobody
Both statutes define a synthetic performer as one who is not recognizable as any identifiable natural person. That clause does more work than it looks like it does, and it runs backwards from how most people assume this kind of law works.
The more anonymous your generated model is, the more certainly she is a synthetic performer and the more certainly you owe a disclosure. If you fine-tune on your actual hired model's face, or the generator produces something recognisably a real person, you have walked out of the synthetic performer definition — and into right-of-publicity and digital replica territory, which is a much more expensive room. There is no configuration in which making the fake person more realistic and more specific reduces your exposure.
This is the same shape as the other lines this industry has been learning all year. Your obligations do not turn on what the frame is, or on how good the picture looks. They turn on a classification: what the object is, what you did to it, whose hands it passed through, and now — with these two statutes — which medium the identical content was delivered in. The eyewear is incidental. The category the asset falls into is everything. It is the same lesson as the one in where your checkout actually renders, arriving from the advertising side.
What the exposure actually is
Subdivision (f) is short: a violation of section 17610 constitutes a violation of section 17500 — California's false advertising statute — and may be enforced under the Unfair Competition Law. That routing matters more than any number, so here is what it actually opens.
A misdemeanour. Section 17500 ends with the sentence that any violation "is a misdemeanor punishable by imprisonment in the county jail not exceeding six months, or by a fine not exceeding two thousand five hundred dollars ($2,500), or by both that imprisonment and fine." Nobody is going to jail over a caption on a frames ad. It is worth knowing the provision says it.
Civil penalties, from public prosecutors only. Section 17206 sets a civil penalty "not to exceed two thousand five hundred dollars ($2,500) for each violation", recoverable in an action brought by the Attorney General, a district attorney, or certain city attorneys. Per violation, on an ad with a lot of impressions, is the part that is not small.
A private route with a real gate on it. Several law firm notes are describing SB 1050 as carrying a private right of action, and comparing it unfavourably to New York's state-only enforcement. That is true as far as it goes, but section 17204 limits private suits to "a person who has suffered injury in fact and has lost money or property as a result of the unfair competition". A consumer who watched an undisclosed AI video ad and bought nothing does not have that. A consumer who bought the frames and says the undisclosed synthetic model is why has a harder argument to make than the headline suggests. Take the private exposure seriously; do not take it at the volume it is being advertised.
What the statute does not settle
Four things, stated as open rather than smoothed over.
Is your own storefront an "advertising medium"? The definition includes "online platform... publisher, or other person or entity that distributes, displays, transmits, or makes available an audio or audiovisual advertisement". A product page video on your own Shopify store is audiovisual and it is displayed to consumers in California. But the duty is to avoid publishing in an advertising medium, and it is not obvious what it means to publish in a medium you are. Paid placement on somebody else's surface is the clear case. Your own product video is not clearly in or clearly out.
What is "prominently"? Undefined. A synthetic model wearing the frames, centre of the shot, for six seconds, is prominent by any reading. A generated pedestrian in the background of a street scene is the contested case and nobody has answered it.
Where does the disclosure go in a six-second ad? "Clear and conspicuous" is defined — "difficult to miss, easily understandable, and presented in a manner that a reasonable consumer would notice" — which is more than New York gives you, and still not a placement rule. On-screen text held long enough to read is the defensible answer. A line in the ad copy below the video is not obviously in the advertisement at all.
Does the still photo stay safe? Outside SB 1050, yes. But New York already reaches it, the largest marketplace in the country already asks you to tag it, and the ordinary false advertising rules never went anywhere. "Not covered by this statute" is not the same as "fine".
Seven things to do before 1 January
1. Inventory the video and audio, not the images. Everyone who did anything about New York audited their photography. Almost nobody audited their video ads, because in June there was no reason to. Start there.
2. Ask the one question per asset, in writing. Was any person in this asset generated? Your agency, freelancer or in-house marketer knows. Nobody has been asked.
3. Add the prescribed phrase to the ones that were. "This performance features a synthetic performer" or "no human performer is depicted". On screen, legible, held long enough to read. For audio, read it.
4. Do not remove your New York captions. They cover an asset class California does not. Removing them because the newer law does not reach stills would be exactly the wrong lesson.
5. Put the indemnity where the creation happens. If an agency generates your creative, say in the contract who owes the disclosure and who pays for the failure. Renewal is the cheap moment for this.
6. Keep the generated people out of anything that reads as a customer. A synthetic person in a testimonial, a review carousel or a "real customers" wall is a federal problem under the FTC's reviews and testimonials rule, independent of both states, and it is a delete-key fix.
7. Leave the product renders alone. A CGI or AI render of the frames with no person in it contains no performer, synthetic or otherwise, and is outside all of this. Do not let an image audit turn into a catalogue rebuild.
The asset class nobody has audited
There is a pattern worth naming here. Each of these rules has landed on the layer of an eyewear store that attracts attention, and none has landed on the layer that converts it.
The face in your video ad now needs a disclosure in two states. The face in your lookbook needs a licence. The face in your virtual try-on collects biometric identifiers and carries per-customer statutory damages if you get the consent wrong. California is simultaneously in court over what you may print on the case, and in Europe the AI Act's transparency duties are working through the same material from a different direction.
Meanwhile the step where single vision becomes progressive, where the index gets upgraded and where the coating gets added — the prescription capture — has no person in it, synthetic or real, and nothing to disclose. It is the least regulated and highest-margin surface in the store.
The disclosure work here is genuinely small: an inventory, a question, and seven words on the videos that need them. It should take an afternoon, and it should not be allowed to take a quarter.
Frequently asked questions
I am not in California. Does SB 1050 reach me?
The advertising medium definition speaks of making an advertisement available "to consumers in this state". For an online store advertising nationally, assume it reaches you. This is the ordinary pattern for state consumer protection statutes, and it is why one state's rule becomes the national floor.
Does it cover the still photos on my product pages?
Not on the face of the statute. An "advertisement" under section 17610 has to be audio, video or audiovisual, and a photograph is none of those — even though a still visual performance does fall inside the definition of a synthetic performer. New York is the state to worry about for stills.
What about an animated still — a slow zoom or a cinemagraph?
That is a video file, and nobody has ruled on whether a photograph with motion applied to it is an "audiovisual message". If you are asking the question, the caption costs less than the answer.
Does an AI voiceover over real footage count?
A synthetic performer is defined to include a "digital figure, voice, or representation". A generated voice that creates the realistic impression of a human performance is within the definition, and audio advertisements are covered in California. This is the exact case New York exempts and California does not.
My model is real but I used AI to change the frames she is wearing. Is that a synthetic performer?
No, provided she remains recognisable as herself — the definition excludes a performer recognisable as an identifiable natural person. But you then need her permission for what you generated, which is a separate and usually larger problem.
Is there a small business exemption?
None was found in the section. The exemptions in subdivision (d) are for expressive works such as films, television and video games where the use is consistent with the work, for generative AI used solely for language translation of a human performer, and for accessibility features. Size is not among them. That is a negative finding from reading the section, not a guarantee.
What happens on 1 January to ads that are already running?
The prohibition is on creating and causing to be published. An evergreen ad that is still being served after the effective date is being made available to consumers then. Treat anything still in rotation on 1 January as in scope rather than grandfathered.
Does the EU require the same thing?
Different mechanism, same direction. The AI Act's Article 50 transparency obligations began applying in August 2026 and are addressed largely to providers and deployers of AI systems rather than to advertisers as such; we cover how that provider and deployer split works for an eyewear store separately. There is no EU synthetic performer statute of this shape.
This article is general information about regulatory developments, not legal advice. SB 1050 has not yet taken effect, no court or agency has interpreted it, and how it applies depends on your channels, your creative and where your customers are. Take advice on your own position before acting. Note also that state advertising law in this area is moving quickly: two states have legislated in ten months, and the statutory text quoted here is the section as enacted, which is the version that governs until it is amended.
VisioncarePro adds structured prescription capture and lens selection to any Shopify product page. It is the step of an eyewear purchase with no performer in it, synthetic or otherwise, and the largest effect on order value. Free to install.
Install VisioncarePro free on the Shopify App Store →
Sources: California SB 1050 (Ashby), "False advertising: synthetic performers", approved by the Governor and chaptered 16 September 2026 as Chapter 246, Statutes of 2026, adding Article 10 (commencing with Section 17610) to Chapter 1 of Part 3 of Division 7 of the Business and Professions Code — bill text read directly, including the definitions at 17610(a)(1), (a)(2)(A), (a)(3) and (a)(6), the prohibition at (b), the prescribed wording at (c), the exemptions at (d), the advertising medium duty at (e) and the enforcement routing at (f); legislative history confirming the final amendment of 19 August 2026, Assembly passage and Senate concurrence on 31 August 2026, enrolment 9 September and chaptering 16 September 2026 · California Business and Professions Code sections 17500 (misdemeanour penalty), 17204 (standing) and 17206 (civil penalty of up to $2,500 per violation) · New York General Business Law § 396-b as amended by S.8420-A / A.8887-B, in force 9 June 2026 · Reed Smith and Davis+Gilbert commentary on SB 1050's effective date of 1 January 2027 and its comparison with New York · FTC Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465.
