California's new AI ad disclosure law arrived after the synthetic performers did
Any video or audio advertisement using an AI-generated performer to sell a product or service in California must now disclose it, under SB 1050, which Governor Gavin Newsom signed at SAG-AFTRA headquarters on September 16, 2026.
Key Takeways
- California's SB 1050, signed by Governor Gavin Newsom on September 16, 2026, requires video and audio advertisements to disclose when they use an AI-generated performer to promote a product or service.
- An advertisement found in violation of California's AI disclosure requirement is barred from continued use, which puts committed media spend and campaign continuity at stake for the advertiser.
- AI-generated performers were already appearing in commercial advertising before any disclosure requirement existed, making SB 1050 a rule written from observation of a practice that had already arrived.
- Most marketing organizations cannot currently say which of their creative assets contain a synthetic performer, and answering that inventory question is the first practical step toward meeting a disclosure rule.
A video or audio ad that uses an AI-generated performer to sell something in California now has to say so. That requirement became law on September 16, 2026, when Governor Gavin Newsom signed SB 1050, authored by State Senator Angelique Ashby of Sacramento, at SAG-AFTRA headquarters. The law covers advertising that promotes a product or a service, and it bars continued use of any ad found in violation of the disclosure requirement.
The part worth sitting with is the order of events. AI-generated performers were already working before SB 1050 existed. Brands were putting synthetic faces and synthetic voices in front of buyers, and California law carried no requirement that the ad admit it. SB 1050 is the rulebook catching up to something that had already arrived.
Every part of this had already been built before a legislature looked at it. The tools to generate a convincing on-camera presenter reached ordinary production budgets, and the ad business used them, because the ad business uses whatever lowers the cost of a shoot. A rule that arrives after a practice has settled has more to describe, because the practice has already taken a shape of its own.
What SB 1050 requires
The requirement is narrow and concrete. An advertisement that uses an AI-generated performer to promote a product or a service has to disclose that. The enforcement piece is what gives the law its weight: an ad found in violation cannot keep running. For an advertiser, that makes disclosure an operating question with a budget attached, because the cost of getting it wrong is the campaign itself.
Newsom signed the bill at SAG-AFTRA headquarters, with National Executive Director Duncan Crabtree-Ireland present. The setting carried the argument.
Californians deserve to know when the person selling them something isn't a person at all.
The people this lands on
Consider the working commercial actor. A face and a voice are the inventory. When a synthetic performer can be generated to do the job a booked performer used to do, the loss shows up as a casting call that never gets posted and a residual that never arrives. Commercial work pays a meaningful share of the income of performers who are not famous, and it is the corner of the business where a generated face is easiest to substitute. SAG-AFTRA has been making this case for years, and SB 1050 answers one piece of it. Disclosure makes the practice visible to the person watching the ad. Visibility is the precondition for everything that comes after, including a market deciding what it wants to buy.
What California built this on
SB 1050 rests on two earlier pieces of California law. The 2024 AI Transparency Act set the state's first broad requirements for identifying AI-generated content. SB 53, in 2025, established safety requirements for frontier AI developers. SB 1050 takes that line of work into one specific commercial setting, the ad break. California now joins New York in requiring disclosure of synthetic performers in advertising specifically, according to the Transparency Coalition, a nonprofit that tracks AI policy.
The same method shows up across the state's recent AI legislation, including California's AI chatbot child safety laws and the state's AI auditor law. Each one names a practice already underway and writes a rule around it. Senator Ashby's bill is a clean example: describe a real use precisely, then require that it be labeled. That is harder legislative work than writing about AI in the abstract, and it produces a rule a compliance team can act on.
What this asks of companies running ads
If your company runs video or audio advertising, the first question is whether anyone can tell you today which of your creative assets contain an AI-generated performer. Most marketing organizations cannot answer quickly, because synthetic elements come in through agencies and production vendors, and no intake form has a field for it. The second question follows from the first: who owns that answer? A disclosure obligation is hard to meet inside a process where nobody is accountable for knowing what is in the finished file.
A contract question sits next to the inventory question. Most agency and production agreements were written before synthetic performers were a line item, so they say nothing about whether a vendor may use one or has to report it. That language is cheap to add at the next renewal and expensive to argue about after a campaign gets pulled off the air.
This is the recurring shape of AI governance right now. The work I do advising the California State University system keeps landing on the same practical point. The first task is an accurate inventory of what an institution is already doing with AI. A rule cannot be applied to a practice nobody has written down. Advertising is that same problem with a shorter deadline and a public-facing artifact at the end of it.
Accountability is the piece that stays with people. A synthetic performer can deliver the line. The decision to run the ad, and the answer when the state asks about it, belong to someone with a name. That is the part of this work I describe as Above the Algorithm: the judgment and the answerability that stay human no matter how much of the output is generated.
The pattern behind the headline
Widen the frame and SB 1050 describes something larger than advertising. A capability arrives and gets used commercially, then the rule is written afterward from observation of what people are already doing. That is the working condition for the next decade everywhere AI touches revenue. Companies that treat each new rule as a surprise will spend that decade reacting to their own operations. Companies that keep an accurate record of where AI sits in their output will find most new rules already answered.
California made the labeling question concrete on September 16. The question underneath it stays open everywhere else: when a rule finally arrives for something your company is already doing, how long does it take you to find out whether you are doing it?
How can a company tell whether its ads use an AI-generated performer?
Start with the asset library and the vendor chain. Ask agencies and production partners to state, per deliverable, whether any on-screen or voiced performance was AI-generated, and record that answer as a field in the asset record instead of an email thread. Post-production houses are a common entry point for synthetic elements, so the attestation should reach them as well. Building this into intake costs far less than auditing a finished campaign under deadline.
What happens to an advertisement that violates California's AI disclosure requirement?
Under SB 1050, an ad found in violation of the disclosure requirement cannot continue to be used. For a brand, that consequence lands on media spend already committed, since a campaign pulled mid-flight loses both the placements bought and the production cost behind them. Planning for it means keeping a compliant disclosure version of each spot ready before launch, which costs little at the editing stage and a great deal afterward.
Which California AI laws came before SB 1050?
Two earlier measures set the foundation. The 2024 AI Transparency Act created the state's first broad requirements around identifying AI-generated content, and SB 53, enacted in 2025, established safety requirements for developers of frontier AI models. SB 1050 narrows that line of work to a single commercial use, advertising, where the audience is a buyer being persuaded. Read together, the three show a legislature working one use at a time, naming a specific practice in each bill.
Here is what makes Alex a credible voice on this topic: Alex Goryachev advises the California State University system on AI governance and shaped a $1.1 billion innovation portfolio across 20 years at Cisco, which is why a disclosure law like SB 1050 reads to him as an operating problem inside the company before it reads as a legal one.
Bringing AI governance questions to your executives or your campus this year? Book a conversation →
.svg)
