The FBI’s Internet Crime Complaint Center says AI-related fraud drained $352 million from Americans age 60 and older in 2025, and much of that money moved because a phone rang with a voice that sounded exactly like a panicked grandchild. The mechanism behind the loss needs only a few seconds of audio pulled from a social media video, then a call timed to trigger fear before anyone has time to check whether the story is true.
How a few seconds of audio becomes a kidnapping call
The scheme follows a familiar shape with an unfamiliar tool attached to it. A caller claims to be a relative who has been kidnapped, arrested or hurt in an accident, and demands money immediately to make the trouble go away. What has changed is the voice on the other end: instead of a stranger doing a rough impression, artificial intelligence now reproduces the exact cadence and tone of the person being impersonated, built from audio the scammer never had to record in person.
The FBI, in a warning covered by Houston television station KPRC, says scammers can build a convincing clone from clips posted to social media or other public sources, then use it to sound “just like” a real family member on a frantic call. Consumer-protection researchers flag the same red flags every time: unusual pauses in the voice, pressure to act before verifying anything, and a request to pay by wire transfer, gift card, cryptocurrency or payment app — all methods that are difficult to trace once the money moves. Cloning tools that once required minutes of clean studio audio can now work from a few seconds of speech pulled off a public video, which is part of why agencies describe the fraud as cheaper to run and harder to spot than the classic “grandparent scam” it evolved from.
Deborah Del Mastro’s $5,000 phone call
Deborah Del Mastro lived the scenario the FBI describes. She received a call from a man claiming he had kidnapped her daughter and was demanding $20,000 for her release. Then she heard what sounded exactly like her daughter’s voice pleading for help, and the fear of the moment overrode her doubts.
Believing her daughter was in real danger, Del Mastro sent the scammers $5,000 before she realized what had happened. The money was never recovered, and she learned only afterward that her daughter had never been in danger at all — the voice on the call had been built from audio, not from a real kidnapping.
The FBI’s $352 million line item
The $352 million figure comes from the FBI’s 2025 IC3 Annual Report, which for the first time in the center’s history devoted a section to artificial intelligence in fraud. According to AARP’s review of the report, the FBI tied $893 million in total losses to AI-related scams last year, including voice cloning used in “family in distress” calls and deepfakes used in investment schemes — and Americans 60 and older accounted for $352 million of that total. The AI figure sits inside a far larger number: Americans of all ages reported more than $20.9 billion in total internet-crime losses to the FBI in 2025, with adults 60 and older alone reporting $7.7 billion of it, a roughly 60 percent jump from the year before.
The report breaks the figure down further. IC3 received more than 3,100 complaints from seniors that referenced AI in 2025, with losses exceeding that $352 million mark, and it counted more than $5 million in losses specifically tied to the “grandparent” or “distress” version of the scam, where a cloned voice claims a loved one is in trouble. Because flagging an AI component on a complaint is voluntary, the bureau’s own data likely understates how much of the underlying fraud actually involved a cloned voice.
Alvaro Puig’s fix: don’t trust the voice, call back
The FTC’s guidance on the scam, written by consumer education specialist Alvaro Puig, comes down to a single habit: verify before sending anything. “Don’t trust the voice,” Puig wrote. “Call the person who supposedly contacted you and verify the story,” using a phone number already known to be theirs rather than any number the caller provides. If that person cannot be reached directly, Puig recommends checking with another family member or friend before acting on the call at all.
The same guidance applies to how the money is requested. A scammer who insists on a wire transfer, cryptocurrency or gift card is choosing a payment method specifically because it is hard to reverse — once the funds clear, there is rarely a way to identify who picked them up or to get them back. That detail, more than any flaw in the cloned voice itself, is what consumer-protection agencies say should end the call. Families that agree in advance on a private code phrase, or simply on the habit of hanging up and calling back on a known number, remove the one thing the scam depends on: a victim who has no time to check.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
More from Morning Overview
