California Now Makes Ads Admit When the Voice Isn’t Human
California has a new rule for advertising: if an AI-generated voice is doing the selling, the ad has to say so. Governor Gavin Newsom signed Senate Bill 1050 on September 16, and the law doesn’t leave voice-over to interpretation. It names the off-camera narrator in so many words.
For anyone who voices commercials, that’s the headline. A synthetic read in a California ad now comes with a label attached, and the label is written in plain English: no human performer is depicted.
What the law actually says
The chaptered text of SB 1050, by Senator Angelique Ashby, adds a new article on “synthetic performers” to California’s false advertising rules. It defines a synthetic performer 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 voice, and note the phrase in whole or in part. A read that’s been partly generated counts.
The label is only required when the synthetic performer appears “prominently,” and the law spells out three ways that happens. The second one is the one voice actors will circle: a synthetic performer “Providing or voicing the on- or off-camera narration or commercial message.” The other two cover a synthetic figure demonstrating the product in the foreground, and one illustrating or reacting to the narration.
When the rule applies, the disclosure has to be “clear and conspicuous,” which the bill defines as “difficult to miss, easily understandable,” and it must use wording “substantially similar to ‘this performance features a synthetic performer’ or ‘no human performer is depicted.'” A breach counts as false advertising under California’s Business and Professions Code, and once a court order is served, a broadcaster, streamer or ad network has to pull the ad and stop taking money to run it.
Who pushed for it, and what they said
Newsom signed the bill at the headquarters of its sponsor, SAG-AFTRA. In the governor’s announcement, he put it simply: “Californians deserve to know when the person selling them something isn’t a person at all.”
Ashby framed it as a jobs bill as much as a consumer one. “The bill requires that synthetic figures be labeled as such, thus protecting consumers from false advertising and disallowing AI to replace real people,” she said in the same release.
SAG-AFTRA’s National Executive Director and Chief Negotiator, Duncan Crabtree-Ireland, said the union “was proud to help craft the language of the bill, which gives consumers greater transparency by making clear when they are seeing synthetics rather than human performers.”
The ad industry fought it
Not everyone cheered. MediaPost’s Wendy Davis reports that the Association of National Advertisers, the American Association of Advertising Agencies, the American Advertising Federation and the Digital Advertising Alliance wrote to Newsom on September 4 urging a veto.
“Rather than targeting deceptive conduct, the bill establishes an overbroad mandatory labeling regime, exposes businesses to opportunistic litigation, and places severe burdens on ordinary commercial speech,” the groups wrote. Their complaint is that the label applies whether or not anyone is actually fooled about the product: “Instead, it mandates disclaimers even when a synthetic performer is used for purely illustrative, background or thematic purposes.”
Santa Clara University law professor Eric Goldman told MediaPost the takedown duty for online publishers may clash with Section 230 of the federal Communications Decency Act. “As a practical matter, I don’t know what publishers would do if they received one of these court orders,” he said. “But in the abstract, not knowing the situation of any individual publisher, it looks like a 230 violation.”
VoiceEditSuiteClean a session in ninety seconds, not three hours.So expect the edges of this law to be argued over. The core of it, a label on an AI read, is the part most likely to stick.
What needs a label, and what doesn’t
A synthetic voice sells it
An AI narrator reads the off-camera script for a California car, bank or grocery ad. An AI figure demos the product, or reacts to the narration. The ad must say something close to “no human performer is depicted.”
Carve-outs
Ads for films, shows and games that use a synthetic character the same way the work does. Ads where AI only translates a human performer’s read into another language, or adds accessibility features. A clone of a recognizable real person falls outside this definition and is handled by other laws.
That translation carve-out deserves a second look if you’re bilingual or you voice spots that get localized. The bill exempts an ad “in which the use of the generative artificial intelligence solely involves the language translation of a human performer.” In other words, an AI-dubbed Spanish version of your English read wouldn’t need a label. Whether the client is allowed to make that version at all is a question for your contract, not this law.
When it starts, and what’s still on the governor’s desk
SB 1050 doesn’t carry an urgency clause, so under California’s usual rule it takes effect on January 1, 2027. Ad campaigns booked for next year are the ones it will touch first.
A companion bill from the same senator is still waiting. SB 1111, which would bring digital replicas into California’s impersonation law, passed the Assembly 77 to 1 and went to the governor on September 8. The Transparency Coalition’s legislative tracker lists it as awaiting his signature, and it notes New York already has comparable ad disclosure requirements on its books. Newsom has until September 30 to act on the bills in front of him.
What it means if you voice ads
First, the limits. The law doesn’t ban AI voices, and it says so outright: nothing in it may be read to “Restrict or prohibit the creation, distribution, or exhibition of synthetic content.” It doesn’t pay anyone, and it doesn’t stop a brand from choosing a synthetic read.
What it does is change the math of that choice. A brand that picks an AI narrator for a California spot now has to tell listeners there’s no human in the room. Plenty of brands spend a lot of money trying to sound warm, local and trustworthy. Some of them will decide that line isn’t the message they want to end on. Nobody knows yet how many.
Three practical things to do now:
- Read the AI and translation clauses in every ad contract. The translation carve-out makes AI-dubbed versions of a human read label-free, so the only protection you have there is what you signed.
- Keep your paperwork. If an ad sounds like you, this isn’t the law you’ll lean on. A voice “recognizable as any identifiable natural person” falls outside its definition, and California’s digital replica and publicity rules take over. Proof of what you recorded, and for whom, is what you’ll need there.
- Use it in conversation. When a producer floats an AI read to save a day’s budget, “you’ll need a synthetic performer disclosure on that” is now a perfectly reasonable thing to say in California.
It’s a small label. But it puts into law something voice actors have been saying for years: listeners care whether a person is talking to them.
Sources: California Legislature, SB 1050 chaptered text and SB 1111 status; Office of the Governor of California, September 16, 2026; MediaPost, September 17, 2026; Transparency Coalition, September 18, 2026.
Lee este artículo en español: California obliga a los anuncios a admitir cuando la voz no es humana
Know something about this story, or spotted an error? Tell the newsroom.