AI Attack Ads Are Reshaping Texas Senate Race

Key Points: A Tight Race, A New Threat

Recent polling places Democratic candidate James Talarico just ahead of Republican incumbent Ken Paxton in their Texas Senate race. The RealClearPolling average shows Talarico at 47.3 percent compared to Paxton's 44.9 percent, a narrow margin that makes this race anything but routine.

However, what makes this contest even more significant is the growing role of artificial intelligence in political advertising—particularly how AI-generated content is being used to attack candidates' personal beliefs and past statements.

• Five of six surveys cited by RealClearPolling show Talarico leading or tied with Paxton.
• Prediction markets are split: Kalshi shows Paxton at 53 percent, while Polymarket displays both candidates at 51 percent.
• Most recently, Paxton released an AI-generated ad featuring a synthetic likeness of Talarico reading from a fictional "King James Talarico Bible," quoting remarks he made in different contexts to paint him as a radical religious figure.

This new form of political advertising has not gone unnoticed. Journalist Mehdi Hasan shared the ad on X, writing, "AI ads like this should be illegal." And political consultant Frank Luntz asked, "Is this the first nationwide TV ad with an AI depiction of a political opponent? I can only imagine what 2028 ads will be like."

Why It Matters: Texas' Republican Legacy and Democratic Ambitions

For decades, Texas has been a stronghold for Republicans, particularly when it comes to federal elections. The last time a Democrat was elected to the U.S. Senate from Texas was in 1988, when Lloyd Bentsen won re-election.

"Texas has not sent a Democrat to the Senate since Lloyd Bentsen won reelection in 1988," I observed while reviewing campaign data. "That history makes Talarico's polling notable, but a narrow survey lead is not the same as a secure advantage in a state where Republican candidates have repeatedly survived competitive races."

What's especially concerning is how AI attack ads might alter that dynamic. In a deeply polarized political environment, where trust in information is already low, these ads blur the line between satire and misinformation—potentially changing public perception before voters have time to fact-check.

The Latest AI Attack: Misleading the Viewer

On September 8, Ken Paxton's official X account posted a 30-second video featuring an AI-generated version of James Talarico. The synthetic character reads from a book titled "The King James Talarico Bible," combining several of Talarico's statements about Christianity, gender roles, and Jesus in a way that never actually occurred together.

View this post on X

The ad captioned: "James Talarico is weaponizing faith to sell a radical agenda, and he even brags about hating Christianity." But several of the remarks referenced in the video were taken from different speeches or interviews. Generative AI assembled them into a continuous scene, creating the illusion of an unbroken performance.

Political commentators and observers quickly reacted. Mehdi Hasan reposted it with a strong call for regulation, while Frank Luntz questioned whether this was the first nationwide ad using AI depictions of opponents—a worrying precedent for future elections.

Polling Points to a Close Race

Multiple polling organizations have attempted to capture voter sentiment in this crucial race:

• Texas Public Opinion Research surveyed 1,000 likely voters from August 21–24 and placed Talarico ahead 48 percent to 42 percent.
• Emerson College Polling found Paxton leading 47 percent to 46 percent in an August 9–10 survey of 1,000 likely voters.
• The University of Texas/Texas Politics Project reported Talarico ahead 42 percent to 39 percent among 1,200 registered voters from August 5–13.

These results reflect the tightness of the race. Despite some variation in margins, none indicate a clear dominance by either side. That makes each campaign message—especially AI-generated ones—more impactful and scrutinized.

What Prediction Markets Show

Prediction markets offer another lens into this election's dynamics:

• Kalshi, a U.S.-regulated platform, gives Paxton a 53 percent chance of victory against Talarico's 47 percent, based on $310,000 in trading volume.
• Polymarket displays both candidates at 51 percent with over $1 million traded.

The divergence between the two platforms is instructive. While one shows a slight edge to Paxton, the other reflects an even contest. These numbers aren't necessarily predictive but rather reflect market sentiment, influenced by liquidity, participant preferences, and contract rules.

Talarico Responds: Shifting Focus to Accountability

In response to Paxton's AI-generated ad, Talarico has turned the tables by focusing on allegations related to Paxton's time in office. His campaign released an advertisement featuring David Maxwell, a former senior aide who reported Paxton to the FBI for alleged corruption.

"Ken Paxton ended my five-decade career in public service because I blew the whistle on his corruption," Maxwell said in a statement accompanying the ad.

This new messaging strategy attempts to counter attacks on Talarico's personal beliefs by highlighting issues of transparency and ethics within the Republican leadership. The campaign argues that voters should be more concerned about past misconduct than present theological debates.

AI Attack Ads Across the Nation

This isn't limited to Texas. AI-generated ads have begun appearing across various political races:

• National Republican Senatorial Committee: Created an AI Talarico reading old posts and giving invented reactions such as "So true."
• Derrick Van Orden: Posted Grok-generated videos resembling Democratic challenger Rebecca Cooke and attributing fictional statements to the character.
• Mike Collins: Released a labeled deepfake depicting Senator Jon Ossoff making remarks about farmers and a government shutdown that Ossoff had not made.
• Republican National Committee: Released an entirely AI-generated advertisement in 2023 depicting hypothetical consequences of another Joe Biden term.

The prevalence of these tools signals a new era in political campaigning—one where digital manipulation can mimic real-world engagement, complicating efforts to assess candidate integrity.

What Happens Next: The Impact of AI on Future Elections

The next few weeks will determine whether the recent AI controversy changes public opinion or simply energizes voters already aligned with one side. For Talarico, holding independents and mobilizing Democrats is essential. Paxton must win over Republicans who remain uncertain about his candidacy.

More broadly, this election highlights a fundamental shift in political communication—where truth can be easily obscured by technology. As artificial intelligence becomes more advanced, so too must our frameworks for evaluating campaign messages. We cannot allow AI to become a tool that erodes democratic discourse rather than enhancing it.