California Deepfake Laws: AI Images, Voice Cloning & Penalties (2026)

California has more deepfake statutes on the books than any other state. As of 2026, state law covers all three major buckets: non-consensual intimate deepfakes carry both criminal penalties and a dedicated civil action; election deepfake restrictions exist but the broadest provision has been permanently enjoined on First Amendment grounds; and right-of-publicity law protects voice and digital likeness against unauthorized AI use for living and deceased performers alike.

Is It Illegal to Make a Deepfake of Someone in California?

It depends on the category of deepfake and what you do with it. California law does not ban the creation of all AI-generated images, but it does impose serious civil and criminal liability in three specific areas: intimate content, election-related content, and commercial use of someone's voice or likeness.

For intimate deepfakes, both creation and distribution can trigger liability. For election deepfakes, the surviving law targets disclosure failures rather than outright prohibition following the 2025 court ruling. For voice and likeness, unauthorized commercial use of an identifiable person's AI replica can ground a right-of-publicity claim regardless of whether the content is sexual or political.

Content that does not fall into these buckets, such as a non-sexual parody video, a fan-made AI music cover that is not sold commercially, or a clearly labeled satirical deepfake, generally is not criminalized under current California law. Federal law fills some remaining gaps, particularly for intimate content involving minors.

Sexual and Intimate Deepfakes

California targets non-consensual intimate deepfakes through two parallel tracks: criminal and civil.

Criminal track: Penal Code § 647(j)(4) covers the non-consensual distribution of intimate images, and the statute expressly defines covered images to include photorealistic and computer-generated depictions of identifiable persons. Conviction is a misdemeanor carrying up to six months in county jail and a fine of up to $1,000. Repeat offenses, or cases where the victim was a minor, raise the maximum to one year in county jail and a $2,000 fine.

Civil track: Civil Code § 1708.86 (enacted by AB 602, effective January 1, 2020, and further amended effective January 1, 2026) creates a dedicated civil cause of action for victims of what the statute calls "digitized sexually explicit material." The statute covers content "created or substantially altered through digitization" that depicts an identifiable individual in the nude or appearing to engage in sexual conduct they did not actually perform. Victims can sue the creator, any person who intentionally distributes the content knowing consent was lacking, operators of services designed to generate such material, and anyone who aids or abets those operators.

Statutory damages run from $1,500 to $50,000, and the statute awards them per work rather than per violation: subdivision (f)(1)(B)(ii) provides for an award of statutory damages "for all unauthorized acts involved in the action, with respect to any one work." Where the defendant acted with malice, the cap rises to $250,000. Victims can also recover actual economic damages, emotional distress damages, punitive damages, attorney fees, and injunctive relief.

Separately, subdivision (f)(2) provides that "a public prosecutor may bring a civil action to enforce this section," recovering civil penalties of $25,000 per violation, or $50,000 per violation where the act was committed with malice. Subdivision (a)(14) defines a public prosecutor as "the Attorney General, a city attorney, county counsel, district attorney, or any other city or county prosecutor," so a local district attorney or city attorney can act without waiting for the state Attorney General.

AI-generated CSAM: California Penal Code § 311.1 covers obscene matter depicting minors, and the statutory definition of "matter" is broad enough to reach AI-generated images. Federal law under 18 U.S.C. § 2256 expressly covers computer-generated images indistinguishable from a real minor and applies regardless of state law gaps.

Election and Political Deepfakes

California has tried harder than almost any other state to regulate AI-generated election content, but the law in this area is in flux after a federal court ruling.

Elections Code § 20010 (AB 730, 2019) is in effect through December 31, 2026. It requires anyone who distributes materially deceptive audio or visual media depicting a candidate within 60 days of an election to include a clear disclosure that the content has been manipulated, and subdivision (c) lets the depicted candidate seek injunctive or other equitable relief plus general or special damages, with attorney's fees available to the prevailing party.

That section repeals itself on January 1, 2027. Subdivision (g) reads: "This section shall remain in effect only until January 1, 2027, and as of that date is repealed, unless a later enacted statute, that is enacted before January 1, 2027, deletes or extends that date." A successor version of § 20010 becomes operative the same day, and it is much narrower. It drops the 60-day window, drops the required "This image, video or audio has been manipulated" disclosure, and drops the "materially deceptive audio or visual media" framing entirely, reverting to a pre-AI rule about campaign material containing "a picture or photograph of a person or persons into which the image of a candidate for public office is superimposed," with a different disclaimer reading "This picture is not an accurate representation of fact." Unless the Legislature extends the current version before then, California's AI-specific election disclosure rule lapses on January 1, 2027, and the guidance in this section needs to be re-checked on that date.

AB 2839 (2024) went significantly further. Signed on September 17, 2024, it banned the distribution of "materially deceptive" AI-generated campaign content, allowing satire and parody only if labeled as manipulated, allowed injunctive relief before and after elections, and gave candidates and elections officials a private right of action. However, a federal judge in the Eastern District of California issued a preliminary injunction against AB 2839 on October 2, 2024, about two weeks after it was signed, and permanently enjoined the law in August 2025 in Kohls v. Bonta. The court held that AB 2839 engaged in content-based and viewpoint-based discrimination by punishing speech only when it could harm a candidate's electoral prospects, while leaving positive deepfake portrayals untouched. The law remains unenforceable unless that ruling is reversed on appeal.

AB 2655 (2024) is California's third enacted deepfake act, codified at Elections Code § 20510 et seq. and titled the "Defending Democracy from Deepfake Deception Act of 2024." It required large online platforms to identify and remove or label materially deceptive election content and gave candidates, elections officials, and prosecutors standing to seek injunctive relief. It is also unenforceable. On August 20, 2025 the same court entered a final judgment declaring AB 2655 preempted by Section 230 of the Communications Decency Act (47 U.S.C. § 230(c)(1), (c)(2)(B) and (e)(3)) as applied to X Corp. and Rumble, and a stipulated order entered on August 28, 2025 bars the state from enforcing AB 2655, in its entirety, against any provider of an interactive computer service. California appealed that judgment to the Ninth Circuit on September 19, 2025 (No. 25-6138), so the question is not finally settled.

The practical upshot: through December 31, 2026, the disclosure requirement of Elections Code § 20010 is the one election-deepfake rule California can actually enforce, and anyone distributing AI-manipulated political content in the state should follow it. The prohibition in AB 2839 and the platform duties in AB 2655 cannot currently be enforced, and § 20010 itself is replaced by the narrower pre-AI version on January 1, 2027. Election-deepfake legislation across the country carries ongoing First Amendment risk; this litigation is the clearest illustration of that.

For a separate take on how California's general AI regulatory framework fits together, see California AI Laws, which covers the broader landscape including SB 942 AI content disclosure requirements.

AI Voice Cloning and Digital Likeness

California's right-of-publicity law is among the most protective in the country and it explicitly covers voice.

Civil Code § 3344 prohibits any person from using another's name, voice, signature, photograph, or likeness for commercial purposes without prior consent. Minimum damages are $750 or actual damages, whichever is greater, plus attorney fees. AI-generated voice clones that are identifiable as a real person and used to sell products, endorse services, or generate revenue without consent fall squarely within this provision.

Compare this to Tennessee's ELVIS Act (Tenn. Code Ann. § 47-25-1101 et seq., eff. July 1, 2024), the national reference law on voice cloning, which created an explicit stand-alone right covering AI voice simulations. California's § 3344 achieves similar results through a broader right-of-publicity framework that has been in place for decades.

Civil Code § 3344.1 (amended by AB 1836, eff. January 1, 2025) extends protection to deceased performers. Anyone who produces or distributes an AI-generated digital replica of a deceased performer's voice or visual likeness without authorization from the estate can face statutory damages of at least $10,000 or actual damages, whichever is greater. The protection lasts for 70 years after death. AB 1836 added specific exemptions for news, documentary, scholarly, satirical, and historical uses, as well as for "fleeting or incidental" appearances.

Labor Code § 927 (AB 2602, eff. January 1, 2025) addresses a different problem: entertainment industry contracts that buried AI waiver language in the fine print. The statute makes a contractual provision unenforceable if it allows an AI-generated digital replica to replace work the performer would have done in person, lacks a reasonably specific description of the intended uses, and was signed without the performer being represented by legal counsel or covered by a collective bargaining agreement addressing digital replica uses. A blanket "we can use your likeness in any medium" clause is no longer sufficient under California law.

These three provisions together make California one of the strongest states for performers and public figures challenging unauthorized AI voice or likeness use. For broader recording and voice consent rules, see California Recording Laws.

Federal Law That Applies in California

Federal statutes layer on top of California law and fill some gaps.

The TAKE IT DOWN Act (Public Law 119-12, signed May 19, 2025) is the first federal law specifically targeting intimate deepfakes. It makes it a federal crime to knowingly publish non-consensual intimate visual depictions of adults or minors, expressly including AI-generated "digital forgeries." Penalties run up to two years in prison (three years for content involving minors). Platforms must remove flagged content within 48 hours of a victim's notice; the compliance deadline for platforms was May 19, 2026. The FTC enforces the platform-removal obligations. TAKE IT DOWN operates alongside California's Penal Code § 647(j)(4) and Civil Code § 1708.86; victims can pursue both federal and state remedies.

The FCC's AI-robocall ruling (FCC 24-17, February 2024) established that AI-generated voices in robocalls are "artificial" under the Telephone Consumer Protection Act. Using an AI voice clone to call phones without prior express consent is illegal nationally. This ruling was triggered in part by an AI-generated fake-Biden robocall in the 2024 New Hampshire primary.

Federal CSAM law (18 U.S.C. § 2256(8)(B)) covers computer-generated images that are, or are indistinguishable from, a real minor. Congress narrowed this definition through the PROTECT Act of 2003 specifically to survive Ashcroft v. Free Speech Coalition (2002), in which the Supreme Court struck down the original, broader ban on virtual child pornography as unconstitutional under the First Amendment. Whether the narrower PROTECT Act standard itself withstands First Amendment scrutiny remains a genuinely contested question in the circuit courts, not a settled one.

Two high-profile proposed federal laws are NOT yet enacted. The DEFIANCE Act (S.1837, 119th Congress) would create a federal civil cause of action for victims of sexual deepfakes; the 118th Congress version passed the Senate but died in the House, and the current version passed the Senate again on January 13, 2026. The NO FAKES Act (S.4591, 119th Congress) would create a federal right of publicity covering AI voice and likeness replicas. Neither has become law as of June 2026. See our coverage of the DEFIANCE Act for the latest.

For AI-specific meeting recording rules in California, the California AI Meeting Recording Laws page covers how state and federal law applies to AI transcription and summary tools.

What Victims Can Do

California gives deepfake victims more tools than most states.

For intimate deepfakes, victims have two immediate paths. First, file a report with local law enforcement or the California Attorney General; Penal Code § 647(j)(4) gives prosecutors a criminal hook. Second, file a civil lawsuit under Civil Code § 1708.86 without needing to prove specific monetary harm: statutory damages start at $1,500 per work even if the financial loss is hard to quantify. Injunctive relief is available to force removal. A public prosecutor, which the statute defines to include the Attorney General, a city attorney, county counsel, a district attorney, or any other city or county prosecutor, can also pursue the creator or platform operator independently.

For election deepfakes, the surviving enforcement avenue through December 31, 2026 is a civil action under Elections Code § 20010, which lets the depicted candidate seek injunctive relief plus general or special damages, with attorney fees available to the prevailing party. That section is repealed on January 1, 2027 and replaced by a narrower version, so confirm its current status before relying on it. The broader AB 2839 private right of action cannot currently be used, and the AB 2655 platform duties cannot be enforced against interactive computer services.

For voice and likeness misuse, a right-of-publicity claim under Civil Code § 3344 (living persons) or § 3344.1 (deceased performers) offers damages and injunctive relief. Performers with AI-waiver concerns in contracts should consult counsel about Labor Code § 927 protections.

Platform takedowns: Under the TAKE IT DOWN Act, victims can submit a removal notice directly to any covered platform, which must act within 48 hours.

California adds its own platform mandate on top of the federal one. Business and Professions Code § 22670 (SB 981, eff. January 1, 2025) defines "sexually explicit digital identity theft" as the posting of "covered material" on a social media platform, meaning an image or video created or altered through digitization that appears to depict an identifiable person's intimate body parts or sexual conduct, reported by the person depicted, who did not consent to it. Section 22671 then requires a social media platform to give California users a reporting mechanism for that material, confirm receipt within 48 hours, send a written status update within seven days, temporarily block the reported material from public view while it investigates, determine within 30 days (extendable to 60 in limited circumstances outside its control) whether there is a reasonable basis to believe the material is sexually explicit digital identity theft, and immediately remove it if there is. That is a statutory duty, not a voluntary platform policy, so a California user whose report is ignored has state law to point to.

For AI voice fraud by phone, complaints can be filed with the FCC and FTC.

Penalties at a Glance

Disclaimer: This page provides general legal information, not legal advice. California deepfake and AI law is changing rapidly: several statutes cited here were enacted or amended within the past two years, the court challenges to the election-deepfake rules are on appeal to the Ninth Circuit, and Elections Code § 20010 as described here is repealed on January 1, 2027, so this page must be re-verified on that date. If you have been harmed by a deepfake or need guidance on compliance, consult a licensed California attorney.

Sources

California deepfake and AI voice law spans multiple statutes and a significant federal court ruling. The site renders citations below from verified primary sources.

Deepfake & AI Voice Cloning Laws by State

California Data Privacy Laws

More California Laws

• California AI Meeting Recording Laws
• California Data Privacy Laws
• California Recording Laws
• California Sexting Laws