Artificial intelligence is making it easier for scammers to clone voices and use them to impersonate relatives, bank representatives, and other trusted people. Worse, criminals don’t necessarily need to build their own technology. Some tools are already being advertised for surprisingly little money.
A study from NordVPN found that commercial activity surrounding AI voice cloning on dark web forums and Telegram channels grew 7.4 times between January-July 2024 and January-July 2026. During the first seven months of 2026 alone, researchers found more voice-cloning mentions than throughout all of 2025.
With October marking Cybersecurity Awareness Month, the findings offer another warning about how AI is changing the way criminals target everyday consumers.
AI Voice-Cloning Scams Are Becoming Easier to Buy
NordVPN analyzed dark web forums and Telegram groups using NordLayer Intelligence, its threat exposure management platform. Researchers tracked posts advertising voice-cloning tools and services, along with discussions about how criminals could use the technology for fraud.
The growth has been steep. According to the study:
- 82% increase: Advertising for voice-cloning tools, services, and pricing rose by 82% in 2025.
- 75% increase: Those advertisements rose another 75% during the first seven months of 2026.
- 7.4 times growth: Commercial activity related to AI voice cloning grew 7.4 times between the two January-to-July periods studied.
- Fraud on the rise: Posts describing specific voice-cloning fraud schemes are on pace to triple from 2025 to 2026.
The research tracks activity among sellers and users of these forums, not the total number of successful scams or the amount of money victims have lost. Still, the numbers point to a growing supply of tools criminals can use to make impersonation attempts more convincing.
Adrianus Warmenhoven, a cybersecurity advisor at NordVPN, said people can no longer assume a familiar voice proves the person on the other end of the phone is genuine.
“Voice used to be the thing that proved you were talking to a real person. That is over, and most people have not adjusted yet,” Warmenhoven said.
That is a troubling shift because phone calls often involve sensitive information, urgent family situations, and financial decisions. A scammer who can imitate a trusted voice may have a better chance of getting someone to react before they stop to question what’s happening.
Real-World Scams Show How Convincing These Calls Can Be
Two recent reports from Florida illustrate how voice impersonation can be used to pressure people into handing over money or trusting a fraudulent caller.
In an October report, ClickOrlando detailed how an Orlando-area woman was targeted in a fake jailhouse emergency.
The woman, identified as Bonnie in the report, received a call from someone claiming to be her friend Carl. The caller said he had been arrested after a crash involving a pregnant woman and pleaded for help. Bonnie later told News 6 that the voice sounded exactly like her friend.
“Exactly like him. Absolutely exactly like him,” she said.
The deception did not stop there. A second caller posed as a public defender and claimed Carl needed money to get out of jail. Bonnie was instructed to prepare cash and paperwork, and a courier came to her home to collect the money.
By the time she contacted the real Carl, she had handed over thousands of dollars. He had not been arrested or involved in the alleged crash.
Bonnie suspected the caller had used AI to clone her friend’s voice. However, ClickOrlando said it had not independently determined how the caller produced the voice she heard.
A separate case involved someone impersonating law enforcement. 404 Media reported on October 19 that scammers allegedly used AI to clone Osceola County Sheriff Steve Moser’s voice. According to the report, the scammers also used the sheriff’s photograph and left a voicemail for a resident, making it appear that the sheriff himself was contacting them.
The report described this as part of a wider pattern of callers impersonating law enforcement. Such schemes can involve claims that someone missed jury duty or has an outstanding arrest warrant. While not every reported police-impersonation call has been confirmed to involve AI, voice cloning adds another way for scammers to make a false identity seem credible.
These cases demonstrate why familiar voices, official-sounding titles, and convincing personal details should not be treated as proof that a call is legitimate.
Criminals Are Targeting Bank Customers With AI Tools
Family emergencies are one concern, but NordVPN’s study found that bank fraud and impersonation of automated bank phone systems were the most discussed voice-cloning use cases in the first seven months of 2026. Celebrity impersonation and cryptocurrency investment scams followed.
Some sellers advertise tools that combine synthetic voices with automated calling systems and other fraud components. These systems can imitate bank phone menus, pose as customer-service representatives, or attempt to trick people into revealing verification codes.
According to NordVPN, some advertised systems are designed to:
- Imitate a bank’s automated phone menu to trick people into entering card numbers, PINs, and other sensitive details.
- Use AI-generated voices in automated calls that ask victims to share one-time verification codes.
- Combine automated calling with phishing websites to collect personal and financial information.
- Impersonate trusted individuals or business executives to make fraudulent requests seem legitimate.
The study also identified tools marketed as AI-powered banker bots, including systems advertised as being able to clone voices and other identifying features. These are claims made by the sellers, not independently verified capabilities.
This is not the first warning about phone-based financial fraud. Cord Cutters News previously reported on the FCC’s crackdown on robocallers pretending to be your bank. That case involved fraudulent calls that spoofed legitimate bank phone numbers to make the callers appear trustworthy.
Now, with the rapid growth of AI-generated voices, another potential layer of deception has been introduced. A familiar-sounding voice combined with a convincing caller ID could make someone less likely to question an unexpected request.
Scammers Are Getting These Tools for Surprisingly Cheap
Perhaps the most unsettling part of the findings is how accessible some of the advertised tools have become.
NordVPN identified several examples of prices listed by criminal sellers:
- Around $30 a week: Saint OTP Bot, a Telegram bot advertised for capturing banking and cryptocurrency one-time codes through automated calls.
- $200 per 1,000 calls: Gorilla’s Call Bot, an AI-voice robocall tool that uses customizable scripts.
- $400 to $500: Berry AI Voice Clone, a standalone voice-cloning product advertised for a one-time payment.
- $10 to $150: Advertised prices for cracked accounts and stolen access to legitimate voice-generation and avatar-creation services.
It should be noted $30-a-week price applies to an automated code-capturing bot and not necessarily a standalone voice-cloning subscription. The other figures reflect separate listings identified by researchers.
NordVPN also found that voice cloning was being incorporated into broader fraud kits instead of always being sold as a separate product. This allows criminals to combine synthetic voices with automated calls, phishing pages, and other tools designed to steal sensitive information.
Some listings involved stolen credentials or unauthorized access to legitimate commercial services. NordVPN said the involvement of those legitimate products in the listings does not mean their providers participated in or approved of the criminal activity.
Meanwhile, the findings fit into a wider pattern of digital threats. A recent study found more than 52.3 billion stolen browser-cookie records, as criminals continue to find ways to exploit technology people use every day.
How to Protect Yourself From AI Voice Scams
Trying to identify a cloned voice by listening for strange pauses, unusual pronunciation, or robotic-sounding speech may not be enough. Instead, the safest approach is to verify the caller’s identity independently.
Warmenhoven has three practical steps to take:
1. Create a family code word.
Choose a private word or phrase that family members can use to verify their identity during an unexpected emergency. If someone claiming to be your child, parent, or another relative calls asking for money, request the code word before acting.
Don’t share it publicly or include it in social media posts.
2. Hang up and call the person back.
If a caller claims to be your bank, hang up and call the number printed on your bank card or listed in its official app. If a relative calls with an emergency, contact them using the number already saved in your phone.
Don’t use a callback number supplied by the caller, and don’t rely on caller ID alone.
3. Watch for urgency and secrecy.
Scammers want you to act before you have time to think. Be suspicious if a caller insists that you send money immediately, share a verification code, or keep an emergency secret from everyone else.
Pause and verify the story through another trusted source before making a payment or sharing personal information.
If you have already sent money or shared sensitive information, contact your bank immediately and report the incident to the FTC at ReportFraud.ftc.gov.
You Can’t Rely on a Familiar Voice Anymore
AI voice cloning has legitimate applications, including helping people who have lost their ability to speak. However, NordVPN’s research shows how criminals are incorporating the technology into ready-made fraud systems.
For consumers, the biggest takeaway is that a familiar voice is not proof of identity. Whether a call appears to come from a bank, a relative, or a business associate, take a moment to verify who is contacting you before handing over money or personal information.
During Cybersecurity Awareness Month, that extra step could help prevent a convincing phone call from turning into a costly scam.
