AI deepfakes pose risks to consumers, voters and others, security experts warn

Advances in artificial intelligence are making it a cinch to create so-called deepfake images, videos and audio recordings, raising concerns that bad actors could bombard people with hyper-realistic scams and spread disinformation. 

Deepfakes that use the voices and likenesses of public figures, including celebrities or political candidates, can also deceive the public in various ways, security experts told CBS News. By one measure, fewer than 10% of people can recognize an audio deepfake, according to Doppel, a social engineering defense platform for detecting AI-enabled threats.

"In a test, only 8% of people realized an AI agent was calling them, whereas 92% of people thought they were speaking to a human," said Doppel co-founder Kevin Tian. 

AI's capacity to churn out deepfakes at scale is a pressing issue in the political arena, according to security experts. AI-generated advertisements featuring deepfakes of politicians have already surfaced across the internet and social media ahead of the midterm elections. 

In one such incident in Texas, a phony, AI-manipulated social video targeting Senate candidate James Talarico purported to show a former student accusing the Democrat of sexual misconduct, according to The Poynter Institute's PolitiFact service.  

"When you can't tell what's legitimate, you question everything," Michael Kaiser, CEO of Defending Digital Campaigns, a nonpartisan organization that provides free cybersecurity services to political campaigns, told CBS News.

As of mid-October, political campaigns had spent more than $137 million on at least 220 AI-generated ads, some of which feature deepfakes of candidates, according to the Wesleyan Media Project, which tracks ads by political candidates.

"AI is making its way into the 2026 midterms in a big way this year. For comparison, we identified less than $50,000 in spending on AI-generated ads in the 2024 cycle," Erika Franklin Fowler, co-director of the Wesleyan Media Project, said in a statement this month. She noted that in most cases, the creators of the ads don't disclose that they were made using AI. 

Even when someone suspects an ad is AI-generated or features a deepfake, the ad's visuals still can make an impression and shape opinion, the project's co-director, Travis Ridout, told CBS News.

"Even if we know it's not real, that sticks in one's mind," he added.

Prompting for deepfakes

Deepfakes are a kind of synthetic media that replicate a private individual's or public figure's image and likeness. Examples include a voice that sounds like a colleague or relative on the other end of a phone line, or a political figure appealing to voters for support or donations. 

Raul Liive, a deepfake expert at Veriff, an AI-powered identity verification company, said the ease and speed with which deepfakes can be made and deployed today pose major risks. 

"Every day that goes by, the generation of deepfake data gets easier and easier," he told CBS News. "If you wanted to fake someone's face years ago, you needed to make an effort to create fake data and use Photoshop. But now, to generate a fake face, you can just prompt your AI tool."

Veriff has documented a 300% increase in deepfake-related fraud attempts against businesses this year, compared with 2025. Liive outlined several examples of how bad actors use the tech to target businesses and unsuspecting consumers. 

"A purported friend of yours calls and asks for a loan, but you know that person, and they've never made that kind of request before," he said.

Other types of scams center around medical emergencies. A caller could say they have a medical bill they urgently need help paying. 

"Ask yourself what is the chance that someone you know would actually ask you something like that?" he said. "You can also hang up the phone and call the contact back on the number you have for them, or use another channel to validate their identity, if the request seems suspicious."

Veriff research on deepfakes found that consumers struggle to spot deepfakes, leaving them vulnerable to manipulation. In one test the firm conducted, 70% of people misidentified a deepfake video of a female character as being real. 

The old telltale signs that something was a deepfake, like a human having six fingers or three hands, are also disappearing from AI content. Veriff found that people look for unreliable visual cues, such as unnatural skin or oddities in a subject's hair or teeth, which AI is increasingly eliminating from the content it generates. 

Meanwhile, eroding public confidence in what they see could also undermine trust in legitimate content. Kaiser pointed to a photo of Sen. Mitch McConnell posing with his wife from his hospital bed, released amid speculation that the senator had died following a medical emergency earlier this year. Although the photo was authentic, some Internet users suggested it had been manipulated.

"That's an example of people just not trusting what they are seeing," Kaiser said. "People don't know if what they are looking at is legitimate, or if requests for donations are real. They're questioning everything that's going on and that disadvantages and erodes trust in the whole system."