Should Texas Require a Label on Every AI-Generated Political Ad?

Texas already has a deepfake election law.

That fact sounds more complete than the law is.

The statute was written in 2019, before synthetic voices, images, and video became ordinary tools in political communication. It creates an offense for making and distributing a deceptive deepfake video within 30 days of an election when the person intends to injure a candidate or influence the result.

That leaves a simpler question in front of voters now.

Should every political advertisement that materially uses artificial intelligence carry a clear label, even when the content is satire and even when it appears more than 30 days before Election Day?

What Texas law says now

The enrolled text of Senate Bill 751 added the deepfake provision to Section 255.004 of the Texas Election Code.

The law covers a “deep fake video” that appears to show a real person performing an action that did not happen. It also requires intent to deceive and limits the offense to publication or distribution within 30 days of an election.

Those boundaries matter.

A synthetic still image is not a video. An altered voice may not fit the definition. A plainly comic AI scene may not have been created with intent to deceive. An ad released 45 days before voting falls outside the 30-day window.

Texas was early. The technology kept moving.

A March 13, 2026 Houston Chronicle report documented the gap between the statute and the current campaign environment. The report described AI-generated political content in Texas races and noted that proposals to remove the time limit and require disclosures did not complete the legislative process.

The strongest case for a label

A disclosure rule does not have to ban the message.

It can let the speaker make the ad and let the voter know how it was made.

That is the cleanest argument for a label. Voters do not need the government to decide whether every joke is funny or every image is fair. They do need a fair chance to distinguish a real recording from a generated scene.

The label can be short:

This ad contains materially AI-generated audio, images, or video.

That sentence does not say the ad is false. It says the media is synthetic.

Supporters would also argue that a universal label is easier to apply than a rule built around proving intent. Intent is hard to establish. The presence of generated media is usually a more concrete question.

The 30-day line also looks increasingly artificial. A false impression can spread for months and become part of the way voters remember a candidate long before the final campaign window opens.

The strongest case against a label

Political speech sits near the center of First Amendment protection.

Any rule written too broadly could burden satire, parody, minor editing, translation, accessibility tools, or ordinary production work.

Does removing background noise with AI require a label? What about automatic captions? What about changing a sky, translating a speaker’s words into Spanish, or creating a clearly absurd image of a candidate on the moon?

If “AI-generated” is not defined carefully, almost every modern ad could trigger the rule. That would turn the label into wallpaper. Voters would stop seeing it.

There is also a grassroots problem. A statewide campaign may have lawyers and compliance staff. A local volunteer making a meme on a phone does not. A rule aimed at professional deception could become a trap for ordinary political expression.

Opponents can reasonably say that fraud, impersonation, defamation, and voter suppression should be punished directly. They do not want a new speech rule that reaches every tool used to make a political point.

Editorial analysis

Texas should require a label when an ad materially generates or alters the voice, face, or actions of a real person.

The word materially matters.

Fixing audio levels is production. Inventing words a candidate never spoke is synthetic political content.

Cropping a photograph is editing. Creating a realistic photograph of an event that never happened is synthetic political content.

The rule should cover paid political advertising and campaign-controlled distribution. It should apply throughout the election cycle, not only during the final 30 days. It should protect clear parody and satire, but the protection should not erase the disclosure.

A joke can carry a label and remain a joke.

The point is not to make AI illegal. Campaigns have used illustration, reenactment, selective editing, and dramatic narration for generations. The point is to restore one piece of context that realistic synthetic media removes.

This is made. It is not a recording of what happened.

What a workable rule needs

Any serious proposal should answer five questions before it becomes law.

1. What counts as material alteration? The rule should focus on a real person’s identity, voice, statements, or actions.
2. Where must the label appear? It should be visible or audible in the ad, not buried on a campaign website.
3. Who is responsible? Paid sponsors and campaign committees are clearer targets than individual users reposting content.
4. What is exempt? Routine editing, accessibility, translation, and obviously nonrealistic graphics need precise treatment.
5. What happens when the rule is broken? Fast correction and disclosure may protect voters better than a penalty imposed after the election.

Texas does not need a ministry of political truth.

It does need rules built for media that can put a real face on an event that never occurred.

The direct question

A label will not make an honest campaign honest. It will not stop every deceptive post. It will not teach every voter to slow down.

It can give people one true fact before they decide what they are looking at.

Should Texas require every paid political ad that materially generates or alters a real person’s voice, face, or actions with AI to carry a clear disclosure label?