Arizona Gave Every Citizen a Lawsuit Against Deepfakers. Let's Talk About That.

Photo by Markus Spiske on Unsplash

Most state deepfake laws focus on one of two things: nonconsensual intimate images or political advertising. Arizona HB 2394, which took effect June 4, 2024, did something broader — it created a general civil cause of action for any person whose digital likeness is used without consent, regardless of the context. That’s a more expansive approach than most states have tried, and it’s worth understanding carefully.

Two Separate Regimes in One Statute

HB 2394 has two distinct parts and it’s important not to conflate them.

Part One: The General Deepfake Provision. Any Arizona citizen — or any person whose likeness is being misused in Arizona — can bring a civil action against anyone who “digitally impersonates” them without their consent. This isn’t limited to sexual content or political advertising. It covers any deepfake of any person for any purpose.

Part Two: The Election-Specific Provision. Political candidates have an additional cause of action during election periods. Within 90 days of an election, creating or distributing a deepfake of a candidate without their authorization is specifically prohibited.

The political provision is the one that made headlines. The general provision is the one with broader impact.

The General Civil Action

Under HB 2394’s general provisions, a person whose digital likeness is used without consent can seek:

• Injunctive relief — get the content taken down
• Actual damages
• If the violation was willful, enhanced damages up to $10,000 per violation

The defendant bears the burden of proving consent. This is significant. In most civil litigation, the plaintiff has to prove the elements of their claim. Here, once the plaintiff shows that a deepfake of them exists and was distributed, the burden shifts to the defendant to show they had authorization.

That burden-shifting is a substantial plaintiff advantage. Creating a deepfake of someone and distributing it means you’d better have documented consent before you do it, because you’ll need to prove it in court.

What “Digital Impersonation” Covers

The statute defines a deepfake as a video, audio, or image that has been manipulated using AI or other technology to make it appear that a real person said or did something they didn’t say or do — where the depiction is realistic enough that a reasonable person would believe it’s authentic.

This is broad. It covers:

• Video deepfakes of someone’s face
• Voice clones of someone’s voice
• Images digitally altered to depict someone in a false context
• AI-generated audio attributed to a real person

Audio-only deepfakes being covered is notable. Most deepfake statutes focus on video. Arizona’s definition sweeps in voice cloning, which makes this law relevant to a wider range of AI applications than the name suggests.

The Election Provisions

During the 90 days before an election, political candidates have an additional cause of action. A candidate can get injunctive relief and damages against anyone who creates, distributes, or publicly displays a deepfake of them with the apparent purpose of influencing the election.

The “apparent purpose” language will be litigated. Parody and satire have long been protected expression under the First Amendment, and courts have generally required that satire be clearly identifiable as such. A deepfake video that appears realistic and is designed to persuade voters is distinguishable from an obvious parody — but the line isn’t always clear.

The First Amendment tension with deepfake restrictions on political speech is real. You have a right to criticize politicians, including through creative means. You don’t have a right to fabricate realistic video of a candidate saying things they didn’t say in order to deceive voters. The statute’s “apparent purpose of influencing” language is designed to capture the deceptive use case without sweeping up protected parody and satire.

Whether it succeeds at that boundary-drawing will depend on how Arizona courts interpret it.

Practical Impact on Synthetic Media Companies

If you’re building products that generate synthetic video, audio, or images of real people:

The consent documentation problem is significant. Under HB 2394, you need affirmative, documented consent before your product can generate realistic synthetic content of a real person. “I agreed to the terms of service” isn’t going to cut it if the terms don’t specifically address synthetic media creation.

Your enterprise customers are also exposed. If a company uses your API to generate synthetic video of real people and distributes it in Arizona, they have liability. Depending on your ToS, so might you, if you knew or should have known what they were doing.

The burden-shift matters. Because the defendant has to prove consent, plaintiffs don’t need to prove you intended harm. You made a realistic depiction of them without authorization. That’s enough to get into court.

The Consent Carve-Outs

The statute has consent and news/documentary exceptions. Journalism, documentaries, news reporting, and similar public interest content that uses synthetic media to illustrate factual events has more protection. Satire and parody — where it’s clear the content isn’t representing itself as real — also has more protection.

These carve-outs are important for media companies and educational content creators. But they require the content to actually be what it claims to be — a parody that looks realistic isn’t protected just because the creator called it parody.

What States Are Converging On

Arizona, Tennessee, and several other states have now passed laws creating civil causes of action for unauthorized use of someone’s digital likeness. The common thread is:

1. The covered person doesn’t have to be a public figure
2. The violation doesn’t have to involve sexual content
3. The defendant bears significant burden once the unauthorized content is shown to exist
4. The remedies include injunctive relief — courts can order content taken down quickly

This is the direction of travel. The synthetic media industry is going to have to build consent-management infrastructure into its products, because the states are building liability around the absence of it.

The law is structurally sound. Realistic AI-generated content of real people without their consent is a genuine harm with no obvious countervailing social benefit. Arizona put a legal remedy in place. It won’t be the last.

You can find the original text of Arizona HB 2394 on the Arizona Legislature’s website.