Deborah Del Mastro, a mother from the San Francisco Bay Area, picked up the phone and heard her daughter sobbing that she’d been kidnapped by a Mexican drug cartel. The voice was unmistakable — the same pitch, the same speech patterns, the same way her daughter said certain words. Del Mastro wired $5,400 across multiple ATMs before she thought to hang up and call her daughter directly. Her daughter was safe. The voice on the phone had never been hers at all — it was an AI-generated clone, built from audio scammers had pulled from somewhere online.
Her case is not an outlier. According to the FBI’s 2025 Internet Crime Report, released in 2026 by the Internet Crime Complaint Center (IC3), Americans filed 22,364 complaints involving artificial intelligence in 2025 — the first year the IC3 tracked AI-enabled fraud as its own category — with total reported losses of approximately $893 million. Voice cloning, fake identification documents, and deepfake videos of public figures were the tools scammers used most.
Key Takeaways
- The FBI’s 2025 IC3 Annual Report recorded 22,364 AI-enabled fraud complaints with roughly $893 million in losses — the first year the agency broke this category out separately.
- Overall internet crime losses hit nearly $21 billion in 2025 across 1,008,597 complaints, up from 859,532 complaints in 2024.
- Americans over 60 lost approximately $7.7 billion to cybercrime in 2025, a 37% increase from 2024 — and older adults are the most frequent targets of AI voice-cloning “family emergency” scams.
- The FBI and American Bankers Association recommend a simple, free defense: a family code word that a scammer’s clone won’t know.
- Voice cloning is also being used against businesses, not just families — the FTC has flagged fraudulent extortion schemes and unauthorized use of professional voice talent as a growing concern.
What the FBI’s Numbers Actually Show
The IC3’s 2025 Annual Report is the most authoritative dataset available on internet-enabled fraud in the United States, compiled from complaints filed directly with the FBI. A few figures from that report are worth separating out, because they answer different questions and get conflated in casual coverage.
| Category | 2025 Figure | Context |
|---|---|---|
| Total internet crime losses | ~$21 billion | 1,008,597 total complaints, up from 859,532 in 2024 |
| AI-enabled fraud (new category) | ~$893 million | 22,364 complaints; first year IC3 tracked this separately |
| Elder fraud (age 60+) | ~$7.7 billion | Up 37% from 2024 |
| Cryptocurrency fraud | >$11 billion | 181,565 complaints |
| Investment fraud | ~49% of all scam losses | Largest single category by dollar volume |
Two things stand out. First, $893 million is very likely an undercount of the true damage, since IC3 relies on victims filing a complaint, and embarrassment keeps many fraud victims — especially older adults — from reporting at all. Second, the AI category and the elder-fraud category overlap heavily: voice-cloning “grandchild in trouble” calls are one of the most common ways AI shows up in a fraud complaint, and older Americans are disproportionately targeted by exactly that script.
How the Scam Actually Works
The mechanics are simpler than most people assume, which is part of why the scam has spread so quickly. Consumer-facing voice-cloning tools have made it possible to generate a convincing synthetic version of someone’s voice from a short public audio sample — a video posted to social media, a voicemail greeting, a podcast appearance, or a clip from a school event. Once a scammer has a clone, the call itself follows a familiar template that predates AI by decades: manufactured urgency, a request for secrecy, and pressure to send money immediately, usually through a channel that’s hard to reverse, like a wire transfer, gift cards, or cash pickup.
What’s changed is not the scam script — it’s the believability of the voice on the other end. A typed scam email is easy to dismiss. A phone call where your child, grandchild, or spouse appears to be crying and in distress bypasses the skepticism most people would otherwise apply.
The threat isn’t limited to families. The FTC has publicly identified two categories of harm from voice-cloning technology: fraudulent extortion schemes targeting families and small businesses, and unauthorized appropriation of a person’s voice that damages professional voice artists and public figures. Security researchers have also documented a rise in voice-cloned “deepfake CEO” calls, where a cloned executive’s voice is used to authorize a wire transfer or override a colleague’s hesitation during a business email compromise attempt — a corporate-fraud evolution of the same underlying technique used against families.
Warning Signs of a Voice-Cloning Scam Call
- Extreme urgency paired with a request for secrecy. Legitimate emergencies rarely require you to tell no one else and act within minutes.
- A request for an irreversible payment method — wire transfer, gift cards, cryptocurrency, or cash handed to a courier.
- A caller who avoids specific, verifiable details about the supposed emergency, or gets vague when asked a question only the real person could answer.
- Background noise or a slightly flat, inconsistent emotional tone that doesn’t quite match the words being said — a common artifact of current voice-cloning tools, though this gap is narrowing as the technology improves.
- A call from an unfamiliar number claiming to belong to a family member, or a caller who says their phone is “broken” or “confiscated” as a reason they’re calling from someone else’s line.
How to Protect Your Family
None of the effective defenses here require new software or a subscription. They’re behavioral, they’re free, and they work regardless of how convincing the technology gets.
Set up a family code word
The FBI and the American Bankers Association both recommend agreeing on a simple phrase in advance, known only to close family members, that anyone can ask for during a suspicious call. A scammer’s clone can copy a voice; it can’t produce a code word it was never told. Pick something ordinary and memorable — not a detail that’s ever been posted publicly.
Hang up and call back on a known number
The single most reliable defense is also the simplest: end the call and dial the person directly using the number already saved in your phone, not a number the caller gives you. As one fraud-prevention expert put it, it’s far better to have a family member laugh at you for checking than to wire money to a stranger based on a voice alone.
Slow the moment down
Every element of these calls is engineered to short-circuit deliberate thinking. Taking even sixty seconds to breathe, hang up, and verify independently defeats the scam’s core mechanism, because the fraud only works while the target is too panicked to check.
Limit how much of your voice is public
Cloning tools generally need only a short audio sample. Setting social media videos to private, being selective about what’s posted publicly, and reviewing older public posts for audio you’d rather not have circulating all reduce the raw material available to a scammer — though this is a mitigation, not a guarantee, since a single public clip is often enough.
Report it, even if you didn’t lose money
Complaints filed with the IC3 at ic3.gov feed directly into the data the FBI uses to track and respond to these schemes. Under-reporting is a known limitation of the current statistics, particularly among older victims who feel embarrassed — filing a report, successful scam or not, helps close that gap.
Voice cloning is one piece of a broader shift in how identity itself is being targeted online. It follows closely behind a 2026 data breach that exposed 153 million driver’s licenses, and it’s part of the same underlying reason the industry has been pushing a shift away from passwords toward stronger, harder-to-fake authentication methods — passwords and even voices, it turns out, are easier to steal or synthesize than most people assumed.
FAQ
How much audio does it take to clone someone’s voice?
Researchers and security firms have widely reported that some consumer voice-cloning tools can produce a usable clone from just a few seconds to a couple of minutes of clear audio, though quality varies by tool. A public social media video, a voicemail greeting, or a recorded interview can all be enough raw material.
How much money have AI voice-cloning scams actually cost people?
The FBI’s 2025 Internet Crime Report recorded roughly $893 million in losses across 22,364 complaints involving AI broadly, which includes voice cloning, fake IDs, and deepfake video. It’s the first year the agency tracked AI-enabled fraud as its own category, and the real total is likely higher due to underreporting.
Are older adults really targeted more often?
Yes. Americans over 60 lost approximately $7.7 billion to cybercrime in 2025, up 37% from the year before, and family-emergency voice-cloning calls are one of the most frequently reported tactics used against this age group.
What should I do if I think I was targeted or scammed?
Stop sending any money immediately, contact your bank or wire service to attempt to reverse the transaction, and file a complaint with the FBI’s Internet Crime Complaint Center at ic3.gov. If a gift card was involved, contact the retailer directly, since some can freeze the card’s balance if reported quickly.
Do businesses face this threat too, or is it mainly a consumer scam?
Both. The FTC has specifically flagged voice cloning as a risk to small businesses through fraudulent extortion schemes, and security researchers have documented cloned-executive-voice calls used to pressure employees into authorizing wire transfers, an AI-era variant of business email compromise fraud.
The Bottom Line
Voice-cloning scams work because they attack trust at its most instinctive level — the sound of a loved one’s voice — and the FBI’s own 2025 numbers confirm this has moved from a theoretical risk to a documented, fast-growing category of fraud. The technology that makes the clone is out of anyone’s control at this point. What isn’t out of your control is the response: a family code word, a habit of hanging up and calling back, and a willingness to pause before sending money under pressure will stop this scam regardless of how realistic the voice on the other end becomes.
This article is for informational purposes only and does not constitute legal, financial, or security advice. If you believe you have been targeted by fraud, contact your financial institution immediately and file a report with the FBI’s Internet Crime Complaint Center at ic3.gov. Figures cited are drawn from the FBI’s 2025 Internet Crime Report, released in 2026, and are current as of September 2026.
