The confident spokesperson in an insurance ad may not exist. Starting January 1, 2027, an advertisement shown to consumers in California may have to say so.
Governor Gavin Newsom signed SB 1050 on September 16 at SAG-AFTRA headquarters. The union sponsored the bill, introduced by state senator Angelique Ashby. The law makes California the second US state after New York to require disclosures when advertising prominently uses a synthetic performer.
The compliance detail matters more than the headline. SB 1050 does not require a label on every ad made with AI. It applies when a realistically human synthetic figure or voice performs a prominent role, and it defines both the threshold and acceptable disclosure language.
This article explains operational implications and is not legal advice. Advertisers should confirm how the law applies to their campaigns with counsel.
What the law requires
SB 1050 defines a synthetic performer as a digital figure, voice or representation created wholly or partly with generative AI that gives the realistic impression of a human performance but is not recognizable as an identifiable natural person.
The distinction is important. A fictional AI spokesperson who looks or sounds human can qualify. A digital replica of a recognizable real person raises a different consent question under California’s AB 2602.
The disclosure rule applies when the synthetic performer is prominent. The statute describes three situations: the performer is in the foreground demonstrating or illustrating a product; provides or voices the on-camera or off-camera message; or illustrates or reacts to that message.
In those cases, the ad must carry a clear and conspicuous disclosure that is difficult to miss, easy to understand and presented so a reasonable consumer would notice and comprehend it in that format. The statute gives two acceptable models: “this performance features a synthetic performer” and “no human performer is depicted.” Wording can be substantially similar rather than identical.
The law covers audio, video and audiovisual messages, including ads distributed online. It excludes synthetic use solely for language translation or accessibility features. It also contains an exception for advertising expressive works such as films, television programs and games when the synthetic performer is used consistently with the work being promoted.
Responsibility begins with the person who creates and causes the ad to be published. Media owners and platforms have a narrower duty: once served with a qualifying court order and enough information to identify a violating ad, they must stop distributing it in California and stop accepting payment for further distribution as soon as commercially reasonable and technically feasible.
How it differs from New York
New York’s synthetic-performer disclosure law took effect in June 2026. California extends the emerging rule into audio advertising, while New York’s framework excludes audio ads.
That difference is operationally significant. California teams need to review AI voices used in radio, streaming audio, podcasts and off-camera narration, not just faces in video creative.
California also sits beside AB 2602, which addresses contracts and consent for digital replicas of identifiable people. A useful production distinction is:
- If the synthetic performance resembles an identifiable real person, examine consent and digital-replica rights.
- If it depicts no identifiable real person but realistically presents a human performer, examine SB 1050’s disclosure requirement.
- If the AI use is not prominent, or only supports translation or accessibility, the new disclosure rule may not apply, but the determination should be documented.
That is a workflow, not a substitute for legal analysis. Edge cases around prominence, realism and distribution should be escalated rather than guessed.
What advertisers should do before January
Start with an inventory of creative already in rotation. Tag every asset that uses a generated face, body, voice, spokesperson, testimonial or narrator. Record where the ad runs, which states can receive it and whether the person is real, fictional or a replica of an identifiable individual.
Next, build the disclosure into production templates. Adding a label after the edit is finished invites placement, duration and legibility problems. Video templates should reserve space and screen time. Audio scripts should include the disclosure in a way an ordinary listener can understand. Short-form formats need the same review as television spots.
Assign an owner for the prominence decision. The law gives useful examples, but teams still need someone to decide whether a background character, reaction shot, product demonstrator or voiceover crosses the threshold. The decision and evidence should live with the approved asset.
Contracts need attention too. Agreements with agencies, production companies, creators and AI vendors should state who identifies synthetic performers, who supplies provenance records, who approves disclosure wording and who pays for remediation if an asset is noncompliant.
Finally, retain the evidence trail. Source files, model settings, performer releases, tool logs and provenance metadata can help a team explain whether an output depicts a real person, a replica or a wholly synthetic performer. Content credentials will not answer every legal question, but they make later review less dependent on memory.
The objection worth taking seriously
TechNet and the Motion Picture Association opposed the bill unless it was amended, arguing that disclosure should focus on material deception rather than apply broadly to synthetic performers.
That concern deserves a fair hearing. A blanket label can impose cost even when an audience is unlikely to be misled, and repeated warnings can become visual or audio clutter. It can also create uncertainty for stylized creative that appears partly realistic.
California’s answer is that consumers have a substantial interest in knowing when an apparent human performance is synthetic. The law narrows the rule through the prominence threshold, the realistic-human definition and specific exemptions. Whether that balance is sufficiently precise will become clearer through enforcement and litigation.
Why APAC teams should care
An agency in Singapore, Jakarta, Manila or Sydney can still create an ad that reaches California consumers. The relevant question is distribution, not where the production team sits.
That makes SB 1050 a briefing and trafficking issue for regional teams handling US campaigns. Creative review needs state-level requirements before an asset is localized, uploaded or handed to a media partner. Teams should also check whether a campaign designed for nationwide US distribution will use a single California-compliant version rather than maintain separate state variants.
With New York and California now setting similar disclosure expectations, many advertisers may adopt the label nationally. A unified standard can be simpler than geofencing multiple versions, especially for connected television, streaming audio and social platforms where distribution can shift quickly.
The law also sharpens a risk already visible in AI video tools. ContentGrip’s guide to Higgsfield notes that synthetic testimonials and realistic AI performers require disclosure and rights review. SB 1050 turns that general warning into a dated compliance project.
The immediate task is not to label every AI-assisted ad. It is to identify the campaigns where a synthetic person is doing the persuasive work, then make the disclosure part of the creative itself before January 1.
This article is produced by ContentGrow. We’re building branded media outlets for B2B companies. Interested in learning more? Learn more.