Louisiana’s first general election under a new political-deepfake law is approaching, requiring campaigns and other speakers to disclose when artificial intelligence materially alters a candidate’s image, voice or likeness in election advertising.

The Legislature passed the rule as HB 459, which Gov. Jeff Landry signed as Act 806. It took effect August 1 and now governs campaign communications ahead of the November 3 election. The measure was authored by Rep. Mandie Landry of New Orleans and passed after a conference committee resolved differences between the House and Senate.

The law does not impose a blanket ban on AI in politics. Instead, it prohibits distributing an election-related communication that uses AI to create or materially alter a candidate’s or recalled official’s image, likeness or voice so it falsely appears authentic unless the communication carries a clear disclosure. That distinction means campaigns may still use synthetic media, parody and production tools, but certain realistic manipulations must be labeled.

A legal analysis by Phelps Dunbar says the statute reaches printed, digital and broadcast communications that expressly support or oppose a candidate, recall or ballot proposition, as well as material that can reasonably be understood as seeking votes in a specific election. Media companies that merely carry paid political content without creating or controlling it are exempt.

Enforcement carries more than a disclosure correction. A person who violates the provisions with intent to injure a candidate’s reputation or deceive voters can face a fine of up to $2,000, imprisonment for as long as two years, or both. Those penalties make the rule consequential for campaigns, consultants, political committees and independent speakers who commission synthetic advertising.

The law follows visible use of AI-manipulated campaign material during Louisiana’s Senate primary. Axios reported that videos depicted Rep. Julia Letlow and then-candidate John Fleming in fabricated scenarios, giving lawmakers a concrete example of how quickly realistic synthetic content can enter a race. The report also noted that disclosure alone may not eliminate deception: research cited by an LSU misinformation specialist found that some viewers still believed a deepfake after being warned.

Louisiana’s approach therefore treats labeling as a transparency tool, not proof that an ad’s underlying claims are accurate. A disclosure tells viewers that AI was used to create or alter the depiction; it does not verify the message, identify every editing technique or substitute for checking the source.

The act sits alongside a separate 2026 law covering AI-generated campaign robocalls. According to the same law summary, Act 559 requires a clear statement at the beginning of a campaign call when AI mimics the voice of a public figure, with the Louisiana Board of Ethics responsible for enforcement.

For voters, the practical signal is straightforward: realistic candidate audio, video or imagery may now carry an AI notice. Its absence does not guarantee authenticity, but its presence confirms that the depicted event or statement was materially generated or changed. For campaigns, the compliance question should be addressed before distribution, because the law applies to the communication itself and not only to the person who appears in it.