When the Candidate Never Said That

Picture the scenario: it’s the final week of a tight primary, and a video starts circulating on social media. A candidate is shown saying something inflammatory, something that would end most campaigns overnight. It spreads fast, gets picked up by local news, becomes the story of the race. Except the candidate never said it. The video was generated, not recorded.

This isn’t a hypothetical. It’s close to what actually happened in a recent Republican Senate primary here in Louisiana, where deepfake-style videos of candidates in fabricated scenarios surfaced during the closing stretch of the race, undisclosed and unlabeled. The backlash was swift enough that it produced an actual piece of legislation within the same legislative cycle.

House Bill 459, sponsored by Representative Mandie Landry, passed and was signed into law as Act 806, taking effect August 1, 2026. For AAF Shreveport-Bossier members, especially anyone who touches political, issue-advocacy, or campaign creative work, this is a law worth understanding in detail, not skimming past as “someone else’s problem.”

What the Law Actually Requires

Act 806 amends Louisiana’s election code, specifically R.S. 18:1463, 1483, and 1501.1, to add a new disclosure requirement around AI-generated and materially altered content in political communications.

The core rule: any electioneering communication that portrays a candidate for office, or an official who is the subject of a recall election, using content created or materially altered by artificial intelligence must include a clear and understandable disclosure stating that AI was used. That disclosure has to follow the same format and placement conventions already required for other political material disclosures under existing law, so it’s not a vague “somewhere on the page” requirement, it slots into a disclosure regime that campaigns and agencies should already be familiar with.

Notice how broadly the trigger is worded. It’s not limited to video, and it’s not limited to material that makes a candidate look bad. Any AI-generated or materially altered image, audio, or video depicting a candidate falls inside the law’s reach, provided the other elements are met.

That’s a meaningfully wider net than “deepfake” alone might suggest to most people, since “materially altered” content can include far more than a fully synthetic video: a manipulated photo, an AI-cleaned-up audio clip, a composited image, any of it could qualify if it changes what the content actually shows the candidate doing or saying.

Who Has to Comply, and What the Real Trigger Is

The statute is written to apply to “any person” making these communications, which is intentionally broad language. It covers candidates and their campaigns, political action committees, individual citizens, and any organization producing electioneering communications, not just professional political consultants or ad agencies. If your agency does creative work for a candidate, a PAC, or an issue campaign, this law reaches your work product directly.

Here’s the detail that matters most for anyone actually producing this content: the criminal penalty attached to the law isn’t a strict liability trap for using AI tools. The penalty, a fine of not more than $2,000, imprisonment of not more than two years, or both, only applies when the undisclosed AI use is done with the intent to injure a candidate’s reputation or to otherwise deceive a voter.

In other words, the law isn’t punishing AI-assisted creative work as such. It’s punishing the combination of AI-generated deception plus an intent to mislead, with the disclosure requirement functioning as the compliance path that keeps legitimate, disclosed AI use out of criminal territory entirely.

That’s an important distinction to sit with, because it means the safest posture for anyone working in this space isn’t “avoid AI tools in political creative,” it’s “disclose clearly, every time, regardless of how minor the alteration might seem.” The intent element gives prosecutors a real hurdle to clear, but it doesn’t give creative teams a reason to get loose with disclosure.

If a piece of content depicting a candidate was touched by AI in a material way and it isn’t disclosed, you’re relying entirely on a future argument about intent to stay out of trouble. That’s not a position any agency should want to be in.

Why This Law Exists, and Why the Timing Matters

Deepfake legislation has been moving through statehouses across the country for a couple of years now, largely in response to the same underlying anxiety: AI tools have gotten good enough, and cheap enough, that a single bad actor with a laptop can produce content indistinguishable from real footage of a public figure. Louisiana wasn’t first to this, but the local incident that preceded HB459’s passage gave it an unusually direct and immediate justification.

This wasn’t a law written in the abstract about a hypothetical future risk. It was written in direct response to something that happened in a real Louisiana race, within the same election cycle the legislature was in session for. That matters for how seriously this should be taken going into future cycles.

Rep. Landry has publicly indicated she expects additional regulation in this space in future legislative sessions, which tells you this is a starting point, not a finished framework. Advertisers and agencies who treat this as a one-time compliance checkbox risk finding themselves behind when the next round of rules arrives, likely with tighter definitions and possibly steeper penalties, given how quickly this first version moved from incident to statute.

The Uncomfortable Part: Disclosure May Not Actually Fix the Problem

Here’s where this gets genuinely interesting for anyone in the advertising and communications field, not just legally interesting but professionally interesting. Researchers who study political misinformation have raised a pointed concern about this exact kind of law: simply disclosing that content was AI-generated may not be enough to correct what a viewer walks away believing.

Studies on this have found that audiences continue to believe fabricated or altered content depicts something real even after being told, clearly and explicitly, that it was AI-generated or altered. The disclosure satisfies the legal requirement. It doesn’t necessarily satisfy the goal the legal requirement was written to achieve.

This puts agencies and creative teams in an unusual spot. You can be in full legal compliance, disclosure present, format correct, and still be participating in a piece of communication that misleads voters about what a candidate actually said or did, simply because the human brain doesn’t fully discount a vivid image or video even when it’s labeled as fake.

The disclosure box isn’t the finish line, and it’s exactly the kind of tension the advertising profession has navigated before in other contexts, the gap between what’s technically compliant and what’s genuinely honest. It’s worth an honest conversation inside your agency about where you personally draw that line, separate from what the statute requires.

What This Means for AAF Members Right Now

If your agency or in-house team does any political, campaign, or issue-advocacy creative work in Louisiana, here’s the practical reality check worth running today:

• Any candidate-depicting content that involved AI tools anywhere in production, generation, retouching, voice cleanup, background replacement, image compositing, needs a disclosure that meets the statutory format, not a buried caption or a barely visible watermark.
• “Materially altered” is doing a lot of work in this statute, and it hasn’t been tested in court yet. Until there’s case law clarifying where the line sits between routine photo editing and a material alteration that triggers this law, the conservative move is to disclose whenever AI touched candidate-depicting content in any way that changes what it shows, rather than trying to argue after the fact that a particular edit was too minor to count.
• The intent requirement means your agency’s internal records matter. If a dispute ever arises about whether AI use was intended to deceive versus a good-faith disclosed creative choice, having a clear internal process, dated files, and evidence that disclosure was standard practice on every piece, not just the ones you were worried about, is your best protection.
• This law is specific to candidates and recall-subject officials. It does not, as written, extend to issue ads that don’t depict a specific candidate, or to commercial advertising generally. Don’t over-apply this law to unrelated creative work, but don’t assume commercial work is permanently exempt from this conversation either, since the trend nationally is toward broader AI-disclosure requirements across advertising categories, not narrower ones.

Questions Worth Sitting With, Wherever You Work Through Them

This is exactly the kind of legislation that benefits from a real conversation among the people who actually do the creative work, not just a compliance memo that gets circulated once and forgotten. A few questions worth putting to your own team this week, whether that happens over coffee, in a Slack thread, or at your next internal creative review:

• Where does your agency draw the line between “AI-assisted” and “AI-generated” for disclosure purposes, and does everyone on your creative team draw it the same way?
• Has your team already used AI tools on political creative this cycle without a formal disclosure process in place, and if so, what would it take to build one before the next election cycle ramps up?
• And given the research suggesting disclosure alone doesn’t fully correct viewer belief, does your agency have an internal standard that goes further than the legal minimum, and should it?

We’d genuinely like to hear how AAF Shreveport-Bossier members are answering these, whether that’s a message to the chapter through our usual channels, a tag on social media, or simply a conversation inside your own shop. This is a genuinely unresolved professional question, not just a legal one, and we’d rather work through it collectively, out loud, before the next big race puts it under a spotlight again.

Our Take

AAF Shreveport-Bossier supports transparency requirements around AI use in political communications. Voters deserve to know when what they’re seeing was generated rather than captured, and a clear disclosure standard gives honest campaigns and agencies a defined, achievable way to stay on the right side of the law. Compliance here is worth treating as an investment in the profession’s credibility, not a burden imposed on it, the same way a cleaner playing field benefits every honest advertiser working it.

At the same time, we think the profession should hold itself to a higher bar than the statutory minimum. A law with an intent requirement and a disclosure escape valve tells you what keeps you out of criminal court. The disclosure box still isn’t the finish line for what keeps your work honest in the eyes of the people who see it. That gap is worth sitting with, and it’s worth continuing to talk about as this area of law keeps evolving, because Rep. Landry has made clear this isn’t the last word from the legislature on AI in political advertising.

This post is intended as a general summary of new legislation for educational purposes and does not constitute legal advice. Members producing political, campaign, or issue-advocacy creative involving AI tools should consult with legal counsel to evaluate specific content against the statute’s requirements, particularly given the untested “materially altered” standard and the intent element attached to the criminal penalty.

Source: Louisiana State Legislature, Act No. 806 / House Bill 459 (2026 Regular Session)