Deepfakes and Synthetic Media: Emerging State Rules on Likeness and Elections
Synthetic media law is not one rule but three clusters, each with its own enforcer and its own remedy. Sorting which cluster a problem belongs to is the first practical step.
Key points
- State synthetic-media laws cluster into election deception, non-consensual intimate imagery, and right of publicity or digital replica protection.
- The federal TAKE IT DOWN Act, enacted in 2025, addresses non-consensual intimate imagery including AI-generated depictions and platform removal duties.
- Election-deception statutes face genuine First Amendment challenges, several are in litigation, and outcomes remain unsettled as of mid-2026.
- There is no general federal right of publicity; it is state law, statutory in some states and common-law in others, with widely varying post-mortem terms.
There is no single deepfake law. Someone dealing with a synthetic video, a cloned voice, or an AI-generated image is in one of three legal clusters, and the answer to "what can I do about this" changes completely depending on which. The clusters are election deception, non-consensual intimate imagery, and right of publicity or digital replica protection. Most rules are state law, with a few federal pieces on top. This is a fast-moving area, and what follows reflects the landscape as of mid-2026.
Three clusters, three different problems
Election deception. A growing number of states regulate synthetic audio and video depicting candidates. The common design requires a clear disclosure, or restricts distribution altogether, within a defined window before an election. Some statutes give the depicted candidate a private right of action; others route enforcement through a state elections agency.
Non-consensual intimate imagery. Most states criminalize distributing intimate images without consent, and many have amended those laws to cover synthetic depictions. Civil remedies often sit alongside the criminal provision, and some states add rules for depictions of minors.
Right of publicity and digital replicas. This cluster covers commercial use of a person's name, voice, image, or likeness. Newer provisions address digital replicas specifically — a synthesized voice or performance standing in for a real person — including protections negotiated for performers.
The clusters overlap. A fake video of a candidate may implicate the election statute, publicity rights, and defamation law at once. But the enforcer, deadline, and remedy differ in each lane, so sorting comes first.
Which lane a problem falls into
| The problem | Likely legal lane | Who usually enforces |
|---|---|---|
| Fabricated video of a candidate near an election | State election-deception statute, plus defamation | State elections agency, or the candidate |
| Synthetic intimate image of a private person | State criminal and civil law, plus federal removal duties | Prosecutors, the depicted person, platforms |
| Cloned voice in an advertisement | State right of publicity, plus federal advertising law | The person depicted, and the FTC |
| AI voice in an unsolicited sales call | Federal telephone consumer protection rules | FCC, state attorneys general |
| Fake endorsement on a product page | Federal deceptive-advertising authority | FTC and state consumer agencies |
| Synthetic replica of a deceased performer | State post-mortem publicity rights | The estate or rights holder |
| Ownership of an AI-generated image | Federal copyright, human-authorship requirement | U.S. Copyright Office, federal courts |
Two cautions. The lanes are defaults, not guarantees — a state may have no statute in a cluster at all. And several of these laws are new enough that no settled case law explains how their terms will be read.
The federal pieces sitting on top
Federal law creates no general deepfake offense, but three threads matter here.
The TAKE IT DOWN Act, enacted in 2025, addresses non-consensual intimate imagery, including AI-generated depictions, and imposes notice-and-removal duties on covered platforms. Deadlines, penalty structures, and agency procedures should be checked against the current official text rather than any secondary summary, including this one.
The FCC has treated AI-generated voices in robocalls as falling within existing telephone consumer protection rules, so a synthetic voice in an unsolicited call does not escape the rules on artificial or prerecorded voices. Guidance and complaint routes are at fcc.gov.
The FTC reaches deceptive commercial uses under its advertising authority. A synthetic endorsement or an AI spokesperson presented as a real customer is a deception problem however the media was produced. The rules sit in the FTC's advertising and marketing guidance, and the same principles drive our piece on endorsement disclosures under the FTC's advertising rules. Impersonation scams aimed at individuals are tracked on the FTC's consumer information site.
Note: The constitutional question is live. Election-speech statutes regulate political expression, which receives the strongest First Amendment protection, and several have drawn challenges on vagueness and overbreadth grounds. Some have been enjoined in whole or in part. Confirm a statute's current status before relying on it: a statute on the books is not necessarily a statute in force.
Likeness, voice, and digital replicas
There is no general federal right of publicity. It is state law, statutory in some states and common-law in others, and the differences are substantial. States vary on what is covered — name, image, voice, signature, mannerisms — on whether commercial use is required, and on how long rights last after death, which ranges from not at all to many decades.
Digital replica provisions are newer. They target synthesized performances and voices used in place of a real person, and several were shaped by entertainment-industry bargaining before appearing in statutes. The recurring elements are consent, specificity about what it covers, and limits on how far a generic release can stretch. Every state also has newsworthiness, commentary, and parody limits — not uniform, and where most publicity disputes are decided.
Who owns a synthetic image
Copyright is federal and answers a different question: not whether the depiction is lawful, but whether anyone owns it. Purely AI-generated material without human authorship is not registrable. The U.S. Copyright Office has addressed that requirement, and how it applies to works containing AI-generated elements, at copyright.gov.
The practical consequence surprises people. Someone who generates a synthetic image of another person may have no copyright in it, while the depicted person may still have publicity or privacy claims. Ownership and legality are separate questions.
Privacy law adds a state layer. California's consumer privacy regime, described on the California Attorney General's CCPA page, is one example of state rules governing personal information that may feed a synthetic depiction — an example, not a national standard. How scraping interacts with it is taken up in our article on automated data collection, access, and copyright.
Preserving proof of a synthetic depiction
Synthetic media disappears. Accounts are deleted, posts removed, and platform copies stripped of the metadata that establishes provenance. Capture the material before reporting it, because a takedown you requested can destroy your own evidence.
- A full-page capture showing the post, account handle, displayed timestamp, and engagement counts.
- The direct URL of the post and of the hosting account, in plain text.
- The original media file at the highest available quality, kept unedited.
- Embedded metadata exported before the file is copied or converted.
- A written log of when and where you first saw it, and how.
- Copies of every platform report filed, with reference numbers and responses.
- Contact details for anyone else who saw it.
- The authentic source material the fake was likely built from, if identifiable.
Preservation matters because authentication is the hard part later. A court will want to know where a file came from and that it has not changed since collection — the problem addressed in our guide to authentication, hearsay, and exhibits in civil cases. If the claim is a false statement of fact rather than a likeness problem, the analysis shifts toward our article on defamation, opinion, and platform issues.
Watch out: Do not record calls or conversations to gather proof without checking the applicable consent rule first. Whether one party's consent is enough varies by state, and some require every participant to agree. Evidence gathered unlawfully creates a second problem on top of the first.
Common questions
Someone posted a fake video of me — where do I actually start?
Preserve first, report second. Capture the post, the URL, the account, and the original file before filing anything, because a successful takedown can erase your evidence. Then sort the problem into a lane: intimate imagery, commercial use of your likeness, or a false factual claim about you. Each has a different route and a different enforcer. Platform reporting runs in parallel with any legal claim and does not replace it.
Are synthetic political ads illegal?
It depends entirely on the state and on timing. Several states require a disclosure on synthetic depictions of candidates within a window before an election, and some restrict distribution outright. Others have nothing. Several such statutes have been challenged on First Amendment grounds and some enjoined, so a statute's presence in the code does not establish it is enforceable now. Check its current status.
Can I register copyright in an image I generated with a text prompt?
Not for the purely machine-generated portions. Federal copyright requires human authorship, and the U.S. Copyright Office has declined registration for material generated without it. Where a human contributed meaningful creative expression — selection, arrangement, substantial modification — that human contribution may be registrable while the AI-generated elements are not. Applicants are expected to disclose AI-generated content. Consult the Office's current guidance before filing.
A cloned version of my voice was used to sell a product. Who handles that?
Probably two bodies at once. Unauthorized commercial use of your voice is a state right-of-publicity claim brought by you, and states differ on whether voice is expressly covered. Separately, presenting a synthetic endorsement as genuine is deceptive advertising, which the FTC can pursue and which state consumer-protection agencies may also reach. The publicity claim compensates you; the advertising enforcement does not, so pursuing both is common.
First moves when synthetic media targets you
- Capture everything before acting. Post, URL, account, original file, metadata, and the date you found it.
- Identify the cluster. Election deception, intimate imagery, or likeness and voice — that decides the deadline and the enforcer.
- Check your own state's statute. Coverage, remedies, and whether a private right of action exists all vary; a neighboring state's rule will not help.
- Report to the platform and keep the record. Removal is often the fastest relief, and the report itself becomes evidence.
- Route the complaint correctly. FCC for synthetic voices in calls, FTC for deceptive commercial use, state agencies for election and privacy issues.
- Involve law enforcement where the conduct is criminal. Intimate imagery and threats belong with prosecutors.
- Re-check the law before relying on it. This area changes between sessions, and injunctions can suspend a statute without repealing it.
Practical step: Set a name and image alert for yourself or your organization, and store captures in one dated folder as you collect them. Provenance evidence is far easier to assemble as you go than to reconstruct months later.
Sources
This is general information, not legal advice. Beacon Legal News is a publication, not a law firm, and reading it creates no attorney–client relationship. Law differs by state and changes; check the linked primary sources or speak with a licensed attorney in your jurisdiction before acting.
Beacon Legal Newsroom
Beacon is an independent legal-information publication. Articles are researched against primary sources and revised when the law moves. How we source · Corrections
Related articles
More in Technology, Privacy & Media →-
Technology, Privacy & Media
Online Reviews and Defamation: Opinion, Fact, Evidence, and Platform Issues
Harsh reviews are usually protected opinion. The line is verifiability, and it decides most disputes long before anyone argues about damages or reputation.
7 min readExplainer -
Technology, Privacy & Media
Terms of Service and Clickwrap: When an Online Agreement Binds
Courts deciding whether online terms bind rarely argue about the clauses. They argue about the screen: what it showed, where the link sat, and what the user had to click.
8 min readAnalysis -
Technology, Privacy & Media
Recording Calls and Meetings: One-Party and All-Party Consent
Federal law lets a participant record a call. A number of states do not. When the two people are in different states, courts have not agreed on which rule wins.
8 min readGuide -
Technology, Privacy & Media
Web Scraping and Automated Data Collection: Access, Contract, and Copyright
Whether an automated collector is lawful is really four separate questions. Federal access law has narrowed, which pushed the fight toward contract, server burden, and copyright.
8 min readAnalysis -
Technology, Privacy & Media
Software Escrow and Source-Code Access in Vendor Agreements
A source-code escrow is only as good as its trigger clause and its deposit. Here is what actually arrives on release day, and what a customer usually discovers is missing.
8 min readExplainer -
Technology, Privacy & Media
Email Marketing and the CAN-SPAM Rules Senders Overlook
The federal email statute is misunderstood in both directions: it demands less consent than marketers think, and it reaches further up the chain than they expect.
7 min readExplainer