Americans lost close to a billion dollars to AI scams in one year – and cybersecurity experts fear this is only the beginning.
Nearly $900 million was stolen in scams that incorporated AI in 2025, according to the first report of its kind from the FBI. The bureau also received more than 22,000 reports about such schemes to its Internet Crime Complaint Center.
One woman in California lost more than $5,000 when a scammer used AI to impersonate her daughter’s voice. Another woman in Ohio lost $1.5 million after fake FBI agents convinced her to drain her bank accounts.
Internet scams are not new; they’ve been around as long as people have been logging on. Fake Nigerian prince emails, phishing schemes and all types of malware have long been digital landmines for people surfing the web. But AI is now changing the cyber con game and costing Americans millions in the process — a number that is only going to grow as AI improves, which it does by the nanosecond.
Michael Machtinger, deputy assistant director of the FBI Cyber Division, told the Wall Street Journal that AI-created fraudulent communications “can look very official and very legitimate to even the most trained individuals.”
‘Today’s AI is the worst AI you will ever use’
As AI companies pitch the public on the urgency of adopting their technology, criminals have been more than willing to heed that advice.
Like everyone else playing around with Claude, Gemini, Grok, and ChatGPT, scammers are still figuring out exactly what they can pull off using the chatbots, according to Jake Braun, executive director of the Cyber Policy Initiative at the University of Chicago’s Harris School of Public Policy.
He told the WSJ that as AI continues to improve, the means by which criminals can use it to bilk people out of their money will likely only get more sophisticated too.
“The AI companies like to say that today’s AI is the worst AI you will ever use. What’s also true is that these are the lowest number of AI complaints we are ever going to see,” he told the paper.
Bob Sullivan, host of AARP podcast The Perfect Scam, explained in March that AI has helped scammers flood the internet with fraudulent offers and malicious schemes.
“We’re getting deluged,” Sullivan said. “A couple of years ago, you might have encountered one or two AI-generated scams a year. Now scammer call centers are sending out tens of thousands of scam messages per minute.”
Wide range of grifts
Consumer protection agencies have collected long lists of all the various ways scammers are using AI to try to rip people off. Both California’s Department of Financial Protection and Innovation and New York City’s Consumer Worker and Protection agency have compiled lists of the methods criminals are using.
A new spin on an old grift is the use of deepfakes to convince people they’re talking to someone they trust — or someone they want to trust like a celebrity or a public figure — in order to convince them to send them money. AI scammers might deepfake a photo or a video of a relative in a tough situation or a celebrity as a means of establishing credibility.
Romance schemes follow a similar tack, using fake images or videos — or even voices — of attractive people or celebrities to convince a victim that they’re interested in them. Once trust has been secured, that’s when the scam hits. The account will then ask the person for money or assistance, and with people’s emotions clouding their judgment, they have been known to fork over thousands of dollars.
While some criminals seek to exploit concern for loved ones or a desire for romance, others appeal to greed.
According to law enforcement agencies, some scammers have created entirely fictional influencers to convince people to invest in fake businesses or to support their non-existent work.
Earlier this year, MAGA influencer Emily Hart — an attractive young woman who espoused far-right talking points online — was revealed to be the creation of a 22-year-old Indian nursing student looking to make a quick buck from ideologues.
The student told Wired that he used Google’s Gemini AI to create the fictional influencer and raked in thousands off the “super dumb” — in his words — MAGA crowd who ate up the rhetoric.
But it’s not just so-called “super dumb” people falling for AI scams. Advice, especially online, has typically been aimed at older Americans who may be less fluent in technology and unaware of the red flags associated with fraud. However, the new FBI report suggests broader messaging may be needed to teach a much younger group of Americans how to keep criminals out of their wallets.
Targeting teens
According to the FBI, teens have become a prime target for scammers in recent years. The bureau reports that it received 31,000 complaints to its crime complaint center from people under the age of 20 last year. That’s up 74 percent from 2024 and is nearly triple the number of complaints it received from the same demographic in 2015.
Ade Clewlow, associate director and senior adviser at cybersecurity consulting firm NCC Group, told the WSJ that teens who have grown up online are more likely to trust what they encounter on the internet and “are just as susceptible as anyone else” to fraud.
Social media sites are especially useful for scammers because they allow them to peer into a target’s network of friends and family and search for vulnerabilities they can exploit.
Focusing on a target’s family was exactly how scammers managed to steal more than $5,000 from Deborah Del Mastro of San Francisco.
Earlier this month, Del Mastro answered a phone call and received terrifying news; a voice on the other end of the call told her they had kidnapped her daughter, Sarah, and demanded a ransom.
She told Good Morning America that the voice told her there was “someone here that you need to talk to” before she heard her daughter’s panicked voice.
“I hear my daughter’s voice — sobbing, trying to breathe, having a panic attack,” Del Mastro said. “And [the voice] says, ‘I’m so sorry, Mom. I am so scared. I’m so sorry.'”
The kidnappers demanded she send $5,400 to multiple locations in Mexico if she wanted to see her daughter again. She obliged, fearing for her daughter’s safety. Once she had hung up with the kidnappers, she called her daughter only to find that she was fine.
Del Mastro realized then she’d been scammed, likely by an AI clone of her daughter’s voice.
Erin West, founder of Operation Shamrock, an organization focused on combating scammers, told KGO that even seconds-long audio clips of a person’s voice — like what might be included in a video shared to Facebook or Instagram — can be used to create believable voice clones.
“What they can do with just a few seconds of your voice [is] they can clone it, and they can essentially produce sound that sounds exactly like you,” she told the broadcaster, adding that AI voice cloning is “only getting worse, and it will only continue to get worse with the use of AI and deepfake technology.”
Another woman in California fell victim to a similar scam. When Abigail — no last name given — received a Facebook message from someone claiming to be Steve Burton, a longtime star of the legendary “General Hospital” soap opera, she readily accepted the message.
The two chatted, moving their conversation from Facebook to WhatsApp. Their conversations continued, and eventually Abigail received a video from the person claiming to be Burton. The man in the video — who looked and sounded just like the soap star — called Abigail “my queen” and promised to love her.
Abigail’s daughter, Vivian Ruvalcaba, told Fox News that she saw the video and warned her mother it was AI, but her mother was convinced it was real.
Before she realized it was all a lie, Abigail had sent the fake Burton $81,000 and sold her condo — at $200,000 under its estimated value — on a promise that she and the faux celebrity would run away together to live in a beach house.
If impersonating loved ones or celebrities isn’t bad enough, fraudsters have also used AI to more convincingly impersonate government workers.
Fake government officials
The FBI’s report found that government impersonation schemes are becoming not only more prevalent but more sophisticated. The bureau’s complaint center received more than 32,000 calls last year about government impersonation schemes, up from around 17,000 in 2024.
“What began as clumsy phone calls from fake IRS agents demanding gift cards has evolved into something far more sophisticated and far harder to dismiss,” Judson Dressler, director of the risk operations center at cyber risk company Resilience, told the WSJ.
Fraudsters use spoofing to make their calls appear legitimate on caller ID systems; lift official seals and logos off agency websites; and employ AI deepfake audio and video to create convincing fakes of public officials, Dressler said.
One 93-year-old woman in Ohio reportedly lost $1.5 million to a government impersonation scam when a fake FBI official convinced the woman that criminals were targeting her assets. They warned her she needed to drain her bank accounts and deposit the money into crypto ATMs for safety. The thieves took the money, with no means for her to recover her savings.
In 2025, Consumer Reports delivered a petition signed by 75,000 Americans to the Federal Trade Commission asking it to hold companies that produce products capable of AI voice cloning accountable for the scams made possible by the technology.
Staying safe in the age of AI
As AI improves, it will become increasingly difficult to know not only what information is real, but whether the people you’re talking to online even exist.
In an effort to better educate the public on the latest methods fraudsters are using, the FBI has bolstered its resources available on IC3.gov.
One way Americans can protect themselves from digital scams is to always verify that the person they’re communicating with is legitimate. If someone calls or messages from a government agency, for example, it’s good practice to call the agency directly to verify the identity of the person.
There’s almost never a time when an agency or company needs immediate payment. If someone contacts you demanding money immediately, that’s a good time to pause and verify, according to the FBI.
According to Navy Federal Credit Union, which released guidance for avoiding AI scams earlier this year, Americans may also want to limit how much of themselves they share online. AI needs very little — a few photos, a few seconds of voice — to create a compelling clone that can be used against your loved ones.
If you suspect you’ve been a victim of a scam, you should act quickly and notify your bank, as well as file a report on IC3.gov. Machtinger told the WSJ that reporting is crucial to stopping fraudsters.
“It could help you and prevent numerous other individuals from falling prey to similar kinds of criminal activity,” he said.
