Artificial intelligence is going to massively transform the political advertising landscape during the 2024 U.S. election cycle—for better or worse.

Political advertising spending is projected to top a record $11 billion while candidates in the presidential and down-ballot races will continue to pour more dollars into digital. Candidates and the ad agencies they hire to handle their fundraising and persuasion campaigns (typically firms handle one or the other) are looking at ways AI can be used on a positive front to help them create more efficient, affordable and targeted digital campaigns. 

But with good use cases, there come bad use cases. Deep fakes have already started to pop up; a fake video circulated on social media last month showing Florida Gov. Ron DeSantis announcing he dropped out of the 2024 presidential race (he has not). The DeSantis campaign also allegedly created fake images of former U.S. President Donald Trump embracing Dr. Anthony Fauci, who led the American response to COVID-19 and has been criticized by DeSantis.

With a current lack of federal regulation on AI in political advertising, the responsibility will fall on the social platforms to self regulate in this area.

“The technological advances in the last 20 years have turned political media on its head and given all of these new tools but also increased risks,” said Tyler Goldberg, North America director of political strategy at Stagwell media agency Assembly. “The platforms are incredibly important for disseminating political information as we shift to being a more digital and online population. There is a responsibility on behalf of the platforms to do things because they are good for democracy.”

Still, there is a call for federal regulation. The Federal Election Committee is looking into regulating AI in political campaigns.

“A private company has different motives and they have different drivers behind their actions,” Goldberg added. “So, it’s important to have the government there to say, at a base level, this is what you need to do.”

Concerns of deep fakes

Misinformation has infiltrated recent U.S. elections. Deep fake videos, generated by advancing AI technology that will make these videos seem very realistic to the average person viewing them, are expected to exacerbate the problem further this election cycle.

Google was the first platform to put guidelines in place for political advertisers using AI. Starting in November, political ads running on any Google platform, including YouTube, that used AI to alter imagery or voices must include a disclosure.

“For years we’ve provided additional levels of transparency for election ads, including ‘paid for by’ disclosures and a publicly available ads library that provides people with more information about the election ads they see on our platforms,” a Google spokesperson said in an email to Ad Age. “Given the growing prevalence of tools that produce synthetic content, we’re expanding our policies a step further to require advertisers to disclose when their election ads include material that’s been digitally altered or generated. This update builds on our existing transparency efforts—it’ll help further support responsible political advertising and provide voters with the information they need to make informed decisions.”

Whether that action will significantly move the needle remains to be seen. Other platforms have not yet released any guidelines pertaining to AI’s use in political ads.

The five political strategists Ad Age talked to for this story said they haven’t heard from any of the other platforms including Meta (parent of Facebook and Instagram), TikTok or X (formerly Twitter) on how they might go about regulating AI to stop the spread of deep fakes on their platforms.

For people trying to release a deep fake video through YouTube, for example, hopefully this regulation at the very least gives them “pause,” said Megan Clasen, partner and co-founder of Gambit Strategies, a digital persuasion agency for Democratic politicians.

“It’s just hard to know whether people will truly be honest about that,” Clasen said.

Many streaming TV platforms, including Hulu, make anyone submitting a political ad include sources from where information is coming. Clasen said she hopes YouTube will do its due diligence in vetting videos to see if AI was used.

More streaming news: See Fox’s election year campaign

“As always, we use a combination of human reviews and other technical tools to examine ads on our platform and determine if the ad was digitally altered or generated,” the Google spokesperson wrote, in response to how it would monitor the use of AI in political ads. “We will disapprove or remove ads we find that violate our policies and may suspend advertiser accounts if we find pervasive or egregious policy violations. Over the years, we’ve developed extensive measures and policies to tackle misinformation on our platforms, including false claims about elections and voting procedures that could undermine trust or participation in democratic processes.”

Realistically, the only way for voters to be truly perceptive to deep fakes are to essentially Google them. “If there's a clip, for example, they show Biden is saying something crazy. Maybe Google and try to see if that clip exists elsewhere,” Clasen said. “I don’t know that you're necessarily going to be able to, with a naked eye, tell that it’s fake.”

While some videos (like the one below) may be obviously fake, others will be harder to detect.

At the same time, it’s not necessarily “reasonable” to expect people to have to double check everything they see online, Clasen said. “People are busy, they have a lot going on, they shouldn't have to be doing their own research every single day to try to determine if something is real or fake.”

A digital strategist working for one of the Republican presidential candidates, who spoke on condition of anonymity, said the best way to determine whether a video is real or fake is by looking at the “mouth to voice consistency; that’s the place it’s hardest to replicate without a human factor.”

On the other hand, the digital strategist said people are starting to use actors behind green screens to create more realistic deep fakes and eliminate some of those consistency issues. The digital strategist said there are a few new generative AI models that are coming out that are going to make “where we are today look like the 1970s,” in a  matter of months, in terms of technological advancements. “It’s terrifying,” this person said.

Also read: 6 AI tools for marketers to know about

Some political advertisers don’t see the government or the platforms being able to keep up with the advancements in AI to properly regulate it.

Rob Shepardson, co-founder and partner of New York-based political and commercial ad agency SS+K, said political advertising regulation has “never been in real time. I don't think [regulation is] going to happen at a speed that’s going to make a big difference in these incredibly fast-moving political campaigns,” he said.

Shepardson worked on former U.S. President Barack Obama’s 2008 and 2012 campaigns. He said he used “big data” for Obama’s successful re-election campaign “but it has just obviously accelerated.”

Regulation in the digital space overall is far behind its time, Assembly’s Goldberg said.

Goldberg said many people are still waiting for social platforms to be required to report political ad spending in the same way TV networks are required—such as having to disclose the entity that is sponsoring an ad (which broadcast ads require but social media ads do not).

“If the government isn’t going to proactively regulate that, like they have not been doing for the past 20 years, the companies are going to be asked to step up,” Goldberg said.

More news: Stagwell and Real Clear Politics partner for ’24

Even Google executives have called for more federal regulation of AI.

“As we’ve said before, AI is too important not to regulate, and too important not to regulate well,” Kent Walker, president of global affairs for Google and its parent Alphabet, wrote in a May blog post. “From Singapore’s AI Verify framework to the UK’s pro-innovation approach to AI regulation to America’s National Institute of Standards & Technology’s AI Risk Management Framework, we’re encouraged to see governments around the world seriously addressing the right policy frameworks for these new technologies, and we look forward to supporting their efforts.”

For now, social media platforms will be looked at to quickly take down deep fakes, but it might not be enough, Shepardson said.

Even if a deep fake video is quickly taken down, it has lasting effects on a campaign, Shepardson said. “People are so prone to believe the worst, particularly in politicians, that if a negative [video] is out there, it doesn't have to be out there for long. The damage is often done.”

“The biggest concern that ... everybody now is worried about is the deep fakes,” Shepardson said.

How AI is enhancing political advertising

Mike Nellis is the founder and CEO of Democratic-focused fundraising and ad agency Authentic. In May, he also founded Quiller—an AI platform that helps Democratic campaigns and progressive organizations draft and deploy fundraising content—where he serves as executive chairman.

Nellis said he founded Quiller to accelerate the work of his team at Authentic with generative AI, which has helped deliver tasks such as coding fundraising emails and creating fundraising buttons. There is currently no tool for creating videos with AI.

“I spoke to a photographer last week who was telling me he was shooting a candidate on a green screen and then using AI to fill in the background,” Nellis said, citing an example. “So, instead of hiring background actors to sit at the kitchen table and pretend like they’re having a real conversation, I just use AI to do that with stock footage or the ability to generate that image.”

As of the time of this writing, Quiller was servicing 75 campaigns and organizations. Although there is no video tool to make deep fakes, Nellis said Quiller ensures no one is doing anything deceitful through the platform because “we have a closed beta, which means we're only letting people in after we vet them and that's on purpose,” he said.

Some believe deep fakes are a minor risk compared to the benefits AI will bring to political advertising, and the strategists who spoke with Ad Age all said they doubted any reputable campaign is out to use this technology for dirty tricks.

If any campaigns were to use AI for deceitful means, Clasen said she would imagine it would be political action committees (PACs). “They are more likely to be a little risky in terms of the messaging that they put out since it's not coming directly from the candidate,” she said.

“The dirty trick issue, while flashy, is overblown,” the digital strategist for the Republican presidential candidate said. “Right now anyone can send a piece of direct mail riddled with outright lies on it, and they could photoshop their opponent into a picture that made them look bad but wasn’t real, but it doesn’t really happen often. Will there be some bad actors? Sure. But there will be in every communication technology, new and old. I think the backlash from someone aggressively using AI to deceive voters in a high profile way would be so bad that it would make the operators as likely to avoid it as the groups themselves.”

This digital strategist said their firm uses AI to create more targeted digital ads. It also uses Open Interpreter, an open-source implementation of ChatGPT Code Interpreter, to do tasks such as writing scripts and scraping videos.

The person said these new AI tools have replaced the cost of “at least three” full-time employees. 

At this point, the agencies are benefiting from these cost efficiencies, the digital strategist said.

“Typically the cost curve works by initially compensating agencies more, and then as agencies compete the commoditization pushes the value to the advertiser,” the person said. “I suspect the same thing will happen here, [it will] probably take a cycle or so to achieve.”

Whether cost-savings—if a campaign, let's say, does the creative in-house and then would reap the cost benefits directly—would allow for more media buys, the digital strategist said: “Eventually, probably, but the more likely scenario is they run more creative and do more testing and experimentation.”

AI will be really powerful in targeting swing voters this election cycle, Shepardson said.

“Getting to donors and optimizing how to extract money from people has gotten much more sophisticated,” Shepardson said. “There’s data that shows outcomes are enhanced with the enhanced ability to speak to [donors] about the issues they care most about, their fears and stuff.”

Swing states including Nevada, Georgia and Arizona, are always where the majority of political advertising is focused. 

Los Angeles is also projected to be an epicenter of political advertising, per a recent report from Stagwell agency Assembly, which estimates political ad spend will reach $12 billion in the U.S. this election cycle. California has a rare Senate seat open with several hopefuls and “at least 10 U.S. House races in California that will be highly competitive in 2024, seven of which are Republican-held seats that Democrats see as viable pickup opportunities,” according to the report.

There may be more use of generative AI to create full ads within down-ballot races, where there is less money to spend.

“You could imagine a scenario where, you know, a state legislator had their creative generated by AI for budgetary reasons,” Clasen said. “You could see a world where races with lower budgets might try to use AI to communicate with voters. But it’s almost like, the more local your race is, it would, to me, be even more risky because there’s even more of a need to be talking to the community because it is so local.”

The digital strategist for the Republican campaign said he believes “AI is going to be ubiquitous enough” that it will be used at all levels this election cycle.

Candidates are going to have to reach voters across all digital platforms—from TikTok, to Reddit, to Instagram, to YouTube, to CTV platforms—and AI is going to be critical in doing that, especially at a local level, said Tal Jacobson, CEO at Perion, a global ad tech company that is working with some U.S. political ad agencies.

“The only way to really deal with all those channels and all those audiences, with all those messages locally, is through AI,” Jacobson said. “You can't generate so much content and so many creative permutations without AI. Even CTVs now can be targeted locally. If you know that in a specific city, not just a state, you're very weak at the polls, you can target them [with] digital out-of-home. The fact that you can strengthen in specific locations, as opposed to nationwide; that’s a lot more powerful."

Nellis said he imagines that most political candidates are experimenting with generative AI, whether they are out there discussing it or not.

Authentic used AI back in 2020 for U.S. President Joe Biden’s campaign. The team developed a chatbot to have AI-generated conversations with “hundreds of thousands” of voters in key battleground states. "We had more conversations with voters that the Biden campaign couldn’t reach,” Nellis said.

“I think there [are] a lot of people who are using AI and ChatGPT and trying to find other tools, but they’re just not talking about it,” Nellis said. “They’ll talk about it one-on-one but they’re not going to go on LinkedIn and talk about it. They’re not going to  get on the phone with the reporter and be like, ‘yes, I’m using it.’”

Because AI is still so new, and with the lack of, but potentially looming, regulations around using it, most political ad agencies and candidates are likely experimenting with AI but not going all in yet.

But that’s likely to change.

Whether voters will embrace AI in political advertising remains to be seen. A recent Pew Research Center survey found that the majority of Americans (52%) say they feel more concerned than excited about the increased use of AI.

Regardless, it’s coming.

“Eventually it will be so commonplace that it won’t matter,” the digital strategist for the Republican presidential campaign said. “The most important thing is whether it seems like the AI is being used to lie to you. We don’t get mad at an ad that uses animation or special effects to tell a story, because we don’t feel lied to. The same thing can be true in political ads: If you’re showing someone something that’s clearly not ‘real’ but is meant to be illustrative, they’ll be fine."