AI-Generated Political Ads Surged in California Elections, Leaving Voters Split

Date:

This election cycle, Californians scrolling through social media or watching campaign ads encountered a strange new reality: candidates battling giant robots, posing as costumed superheroes, or cast as gladiators performing for roaring crowds. None of it was real. All of it was generated by artificial intelligence, part of a wave of AI-produced political advertising that swept through races up and down the state.

Nationally, campaigns had spent more than $137 million on AI-generated ads by mid-October, according to the Wesleyan Media Project, with Republican-aligned groups responsible for roughly 80% of that spending. In California, the trend showed up in races big and small. Former Los Angeles mayoral candidate Spencer Pratt used AI to cast himself as Batman. Steve Hilton, who ran unsuccessfully for governor, appeared in an AI-generated spot as a gladiator triumphing before a packed coliseum that included rivals Xavier Becerra and Gov. Gavin Newsom.

Not every example was played for laughs. In a San Diego-area school board race, a mailer used AI imagery to show a male candidate working confidently behind a desk while his female opponent was depicted running a bake sale — an image she condemned as sexist and misleading.

To understand how these AI-generated visuals are reshaping political messaging — and what risks they pose to voters — we spoke with Ramesh Srinivasan, a UCLA professor of information studies and host of the Utopias podcast. His responses have been edited for length and clarity.

Q: It feels like AI-generated political content is everywhere this year. Is that accurate, or am I just noticing it more?

A: It’s absolutely true. We’re seeing no shortage of what people call “AI slop” — content that’s clearly artificial and not meant to deceive anyone. But the more serious issue isn’t the obvious stuff. It’s how AI is being used to create highly targeted, almost propaganda-like messaging aimed at individual voters.

A decade ago, there was major controversy over Cambridge Analytica, a firm that used people’s personal data to predict their psychology and craft messaging designed to influence not just who they voted for, but whether they voted at all. That concern was never fully resolved, and now AI has made it exponentially easier. Campaigns no longer need to go through the laborious process of buying data, analyzing it and building tailored content around it. AI systems can generate that content almost instantly.

Q: So you’re saying algorithms can essentially deliver a custom political ad to an individual voter, tailored to their psychology?

A: Exactly. If someone has access to your data — not just what you post on Instagram, but things like your credit card records — they can use AI to generate content specifically designed to influence you. That’s part of why this is so concerning. There’s often no disclosure telling people whether what they’re viewing was made by a human or a machine. I’ve been working with members of Congress and California’s state Senate to push for mandatory watermarking of AI-generated content.

Q: What exactly does watermarking mean, and why does it matter?

A: A watermark is a permanent marker embedded in an image or video that identifies it as AI-generated — something that can’t be erased or edited out. It’s essentially a transparency label. Much of the AI-generated political content circulating on platforms like TikTok and Instagram during this election cycle carried no such disclosure.

Q: Some of Steve Hilton’s AI-generated ads did include text acknowledging they were AI-created.

A: Several campaigns have started doing that voluntarily. But the larger issue is that voters need a reliable way to know what’s real and what isn’t, especially since so many of us now experience public life primarily through our screens.

Q: Are voters actually being fooled by these AI-generated images?

A: In many cases, it’s genuinely impossible to tell the difference — and that’s not voters’ fault. There’s a difference between obvious AI content, like when the president posted an AI-generated image of himself resembling a religious healing scene, with a patient who oddly resembled “Daily Show” host Jon Stewart. Everyone recognized that as AI; no one was trying to deceive anyone there. That’s different from AI content specifically engineered to pass as authentic.

Q: Does AI-generated content actually change how people vote?

A: It comes down to trust and perceived authenticity. Early research suggests people generally can’t distinguish well-made AI content from real footage or images. And if what we see shapes how — or whether — we vote, then that’s a serious concern for democracy.

This isn’t only about turnout. It’s about whether people share any common understanding of reality. Social media’s ability to micro-target individuals with different versions of “the truth” was already eroding that shared foundation. AI is accelerating the problem. Data is being used to shape what we see, what we believe and how we feel, and when that happens at scale, it becomes extremely difficult to have any kind of collective governance or basic agreement on facts.

Q: Is this distrust showing up across the political spectrum?

A: Very much so. During the Biden administration, there were accusations — accurate or not — that large tech companies were suppressing certain viewpoints. Now we’ve clearly seen many of those same companies align closely with the current administration; that was evident at the inauguration. We’ll see how that evolves after the midterms.

Technology should expand our capabilities, not replace our judgment or our agency. AI tools can help us explore ideas and understand complex issues, but they shouldn’t be the final word. If AI systems are instead used to replace human decision-making, human livelihoods or human perspectives, then people lose their place in the democratic process altogether.

The real question moving forward is whether our data is being harvested to serve us, or to manipulate us. Can we still encounter the world in a shared, trustworthy way? These questions are still open, and the answers depend on real regulatory action — requiring disclosure when personal data is collected, compensating people for that data, and mandating watermarks on AI-generated content.

Regulation isn’t about stifling innovation. It’s about steering it responsibly, ensuring that technological progress moves forward alongside social and democratic values — not at their expense. Technology is never created in a vacuum; it reflects the values we choose to build into it.

Q: Looking ahead, what role do you expect AI to play in the 2028 elections?

A: The key question is whether leaders will emerge who understand that these technologies are now permanent fixtures of our political landscape — and who commit to using them responsibly. AI can be a powerful tool for helping people understand the world more deeply, not a replacement for human judgment. Getting that balance right will be essential to protecting our democracy going forward.

Original source: CalMatters

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Share post:

Subscribe

spot_imgspot_img

Popular

More like this
Related

As El Niño Approaches, a Flood-Prone Town’s Pioneering Insurance Program Collapses

As another El Niño winter bears down on California,...

California Lawmakers Now Represent the State’s Most Diverse Legislature in History

Southern California voters helped usher in the most diverse...

Voters Reject Proposition 4, Blocking Public Funding of Political Campaigns in California

California voters have rejected Proposition 4, leaving in place...