Deepfakes and the Law

Deepfake law in the United States is a patchwork: one federal law aimed at intimate images, a state election law here, a state criminal law there, and a thicket of platform rules that change faster than any legislature. This lesson covers what applies to anyone posting, campaigning or advocating in Idaho, as of October 2026. It is general education, with every statute linked so you can read it yourself.

The map

Idaho's election deepfake law

Idaho Code 67-6628A, named the Freedom From AI-Rigged (FAIR) Elections Act, took effect in 2024.

The statute defines synthetic media as an audio or video recording of a person's speech or conduct (still images are not covered), made with generative adversarial network techniques or other digital tools, that is realistic but false. Two tests: it would appear to a reasonable person to show something that did not happen, and it gives a very different impression than the unaltered original would.

What it does:

• A candidate whose speech or conduct is deceptively shown through synthetic media in an electioneering communication may seek an injunction and sue the maker (the "information content provider") for general and special damages. The winning side can recover attorney's fees and costs.
• The candidate must prove the synthetic media by clear and convincing evidence.
• Including the statute's disclosure is an affirmative defense.
• Outlets that merely carry the content are generally not liable, unless they remove the disclosure or alter content into synthetic media. Broadcasters carrying ads under 47 U.S.C. 315 are excluded.

The disclosure, exactly

The wording is: "This (video/audio) has been manipulated", using video or audio as fits.

• Video: prominently displayed text, easily readable by the average viewer, for the entire length of the video.
• Audio only: read clearly, at a pitch the average listener can easily hear, at the beginning and the end, and repeated at intervals of no more than two minutes if the audio runs longer than two minutes.

One open question, carried over from the advertising module: 67-6628A does not define "electioneering communication." Idaho's campaign finance definitions in 67-6602 use a narrow version tied to 30 days before a primary and 60 days before a general election. Whether that narrows the deepfake law appears unsettled. Nobody should count on the gap.

Idaho's explicit synthetic media crime

Idaho Code 18-6606 makes it a crime to knowingly disclose explicit synthetic media of an identifiable person when the discloser knows or should know the person did not consent and it would cause substantial emotional distress, or does it to harass, humiliate or degrade, or threatens disclosure to extort money. It is a misdemeanor, and a felony (up to 10 years and a $25,000 fine) if the person depicted is under 18 or the offender has a similar conviction within the past five years.

The federal TAKE IT DOWN Act

Signed May 19, 2025 (Public Law 119-12), the TAKE IT DOWN Act has two halves.

The crime

It added subsection (h) to 47 U.S.C. 223. Knowingly publishing an intimate image of an identifiable person without consent through an interactive computer service is a federal crime when the statute's conditions are met, and the law covers both authentic images and digital forgeries: AI-made or altered images a reasonable viewer cannot tell from real.

The statute says consent to create an image is not consent to publish it. It has exceptions for law enforcement, good-faith disclosures such as reporting unlawful conduct or seeking help, and a person sharing images of themselves.

The takedown duty

Covered platforms (sites and apps that mainly host user content) had one year to build a notice-and-removal process. The FTC announced it began enforcing on May 19, 2026. Platforms must remove reported images, and known identical copies, within 48 hours of a valid request. A law firm summary puts civil penalties at up to $53,088 per violation, a cap the FTC adjusts for inflation each year.

The debate is real and worth knowing. Victim advocates pushed hard for the law. Free speech groups, including the Electronic Frontier Foundation, the Center for Democracy and Technology and the Freedom of the Press Foundation, warned that the takedown process has no strong penalty for false requests and could be used to remove lawful content, including journalism. Both concerns can be true at once. The first enforcement cases will show which one wins out.

Platform rules as of October 2026

YouTube does not require disclosure for clearly unrealistic content (animation, fantasy), for beauty filters, captions or color fixes, or for cloning your own voice for a voiceover. A disclosed video is not restricted or demonetized for that reason alone, YouTube says. Creators who repeatedly skip disclosure can lose content or Partner Program status.

What is not law yet

The NO FAKES Act, which would create a federal right against unauthorized digital replicas of a person's voice or likeness, cleared the Senate Judiciary Committee unanimously in June 2026. It had not passed the full Senate as of October 2026. The FCC proposed AI disclosure rules for broadcast political ads in 2024; the advertising module and the FCC's docket carry the current status, and no federal broadcast rule should be assumed until it is final.