AI Voice Cloning Scams: How Scammers Fake a Loved One's Voice

AI voice cloning allows scammers to create fake voices that sound like a parent, child, or spouse. They use these cloned voices to make urgent requests for money transfers, gift cards, or wire payments. This type of scam is becoming more common, with FBI data showing the growing use of AI in reported fraud.

In 2025, the FBI's Internet Crime Complaint Center logged over 22,000 complaints involving AI, with nearly $900 million in reported losses, and voice cloning is one of the tactics behind that surge.

This pattern matches a 2026 FBI Internet Crime Complaint Center advisory, which found that criminals combine AI-generated voice cloning with fake videos and cloned websites to make fraudulent requests appear legitimate, and recommend verifying any unexpected or urgent request through a separate, already-known contact channel before sending money or personal information.

This guide explains how organizations can protect themselves from deepfake vishing attacks, including AI voice cloning and phishing.

How Do AI Voice Cloning Scams Happen?

An AI voice cloning scam works by feeding a short audio sample of a real person's voice into an AI tool, which generates new speech in that same voice reading a script the scammer controls. The FBI's Internet Crime Complaint Center (IC3) confirms this method is now an established fraud tactic, not a fringe technique.

Here are the steps that usually show how AI voice scams happen:

• Sample collection: The scammer gathers a short clip of the target's voice, usually pulled from public content such as social media videos, voicemail greetings, or recorded calls.
• Clone generation: The audio sample is fed into an AI voice cloning tool, which learns the person's vocal patterns and produces a synthetic version of their voice.
• Script creation: The scammer writes a message for the cloned voice to "say", typically a short clip impersonating a close relative in a crisis situation, asking for immediate financial assistance or demanding a ransom.
• Delivery: The cloned audio is played over a phone call or voicemail, often paired with a spoofed caller ID showing a familiar number.
• The ask: The message pressures the recipient into sending money quickly, usually through cash, wire transfer, gift cards, or cryptocurrency, before the target has time to verify.

If you've already been targeted by one of these calls, Unscammed can help you file reports with the right agencies in one pass, instead of tracking each step yourself.

Why Are Voice Cloning Scams Becoming More Common?

Voice cloning scams are growing mainly because the tools have become more accessible. What once required technical skill, paid software, and several minutes of clean audio can now produce a usable clone from just a few seconds of speech, often through free, browser-based tools that need no special expertise.

People also generate large amounts of public voice data without realizing it, through social media videos, podcast clips, and voicemail greetings, giving scammers easy source material without needing direct contact with the target. The process itself has also gotten faster, with a finished clone now achievable in well under an hour.

According to the FBI's Internet Crime Complaint Center (IC3), AI was added as a formal crime category for the first time in its 2025 Internet Crime Report, logging 22,364 complaints and nearly $893 million in losses.

How Do Scammers Get Audio Samples, and How Much Do They Need?

According to the FTC, scammers only need a short audio clip of a person's voice, often pulled from content already posted online, along with a voice-cloning program to generate a convincing fake call. The FBI's Internet Crime Complaint Center (IC3) confirms the same pattern, noting that criminals generate short audio clips containing a loved one's voice specifically to impersonate a close relative in a crisis situation.

Where scammers source audio:

• Social media videos where a person's voice is audible, even briefly
• Voicemail greetings, which offer a short, clean sample with no direct contact needed
• Public appearances such as interviews, webinars, podcasts, or recorded calls
• Voice notes shared through messaging apps that are later leaked or forwarded

Both agencies point to the same underlying issue: the audio doesn't need to be obtained directly from the target. Anything posted publicly, even in passing, can be extracted and reused. This is why the FBI's guidance focuses less on avoiding phone calls altogether and more on limiting how much voice content is available online in the first place.

What the FBI recommends to reduce exposure:

• Limit online content of your image or voice
• Make social media accounts private
• Restrict followers to people you know

The amount of audio required to begin the cloning process is minimal. Neither agency specifies an exact duration, but both describe it as a short clip, meaning scammers don't need extended recordings or repeated access to someone's voice. A single public video, greeting, or voice note is treated as sufficient starting material.

AI Text-to-Speech vs. Voice Cloning vs. AI Voice Changers

These three technologies are often confused because they all alter or generate speech, but each works differently and carries a different level of risk when used in a scam. Understanding the distinction helps clarify why voice cloning specifically is the version tied to impersonation fraud.

Technology	What It Does	Tied to a Specific Person?	Common Scam Use
Text-to-Speech (TTS)	Converts written text into spoken audio using a generic or stock voice.	No	Robocalls, automated phishing messages
Voice Cloning	Analyzes a real person's voice sample to generate new speech that sounds like them.	Yes	Fake emergency calls, executive impersonation, bank account access
Voice Changer	Alters pitch, tone, or characteristics of a live voice in real time during a call.	No. It modifies the speaker's own voice, not a copy of someone else's.	Disguising identity during a live scam call

How Can I Tell If a "Family Member's" Voice Is Fake?

A familiar voice can sound convincing, but the voice itself is not enough to confirm who is calling. Look at the caller's behaviour, the details they provide, and what they are asking you to do. These checks can help identify a possible impersonation scam.

1. Immediate pressure to act: Be cautious if the caller insists that you must send money or take action immediately and does not give you time to verify the situation.
2. Resistance to a callback: Try calling the family member back using a phone number you already have saved. If the caller tells you not to call back or claims their usual number cannot be reached, verify the situation through another trusted contact.
3. Vague or unusual details: Pay attention if the caller cannot answer simple questions or provides details that do not match what you know about the person.
4. Unusual payment requests: Requests for cryptocurrency, gift cards, wire transfers or other difficult-to-reverse payment methods are common warning signs in impersonation scams.
5. A familiar caller ID: Do not assume a call is genuine because the number looks familiar. Caller ID information can be spoofed, so verify the person's identity independently.

What's a Safe Word Families Can Use?

Both the FBI and FTC recommend establishing a codeword or phrase with family members in advance, specifically as a countermeasure to AI scams. IC3 lists this as a direct tip to protect yourself: create a secret word or phrase with your family to verify their identity.

How to set one up, based on FBI and FTC guidance:

• Choose a word or phrase that isn't tied to public information about the family (not a pet's name posted on social media, for example).
• Share it in person or through a private channel, not over text or social platforms where it could be exposed.
• Agree that the word must be used any time a call, text, or message involves a request for money or urgent help.
• Treat refusal or inability to provide the word as a reason to hang up and verify independently.

What Tools Exist to Detect AI-Generated Voices?

Several tools are available to help identify whether a voice on a call or recording has been AI-generated. These fall into two categories: consumer-facing apps built for individuals, and enterprise-grade systems used by call centers and financial institutions to screen calls at scale.

Tool	Type	What It Does
McAfee Deepfake Detector	Consumer	Scans audio in videos to flag AI-generated speech; built into select McAfee security products.
Pindrop	Enterprise	Analyzes call audio in real time to detect synthetic voice and caller fraud in contact centers.
Resemble Detect	Enterprise	Screens audio for signs of AI voice generation, used by businesses handling high call volumes.
AI Voice Detector	Consumer	Standalone tool for checking whether a specific audio clip is AI-generated.

Can Businesses Be Targeted Too?

Yes. The FBI's Internet Crime Complaint Center (IC3) tracks this as a business email compromise (BEC) variant, where a cloned voice is used to authorize a fraudulent wire transfer or gain account access rather than to fake a family emergency. IC3 states that criminals also use AI-generated audio clips of individuals to obtain access to bank accounts, extending the same technique beyond personal scams into a financial fraud vector aimed at organizations.

Metric	Figure
BEC losses	$3.046 billion
Total AI-related fraud losses	~$893 million
Common target roles	CFOs, finance/accounts payable staff
Typical method	Cloned executive voice + spoofed caller ID, paired with email instructions
Reporting channel	ic3.gov

Could your team spot an AI voice scam? AI-generated calls can make it harder for employees to know whether a request is genuine. Unscammed’s business scam protection platform helps small businesses train employees with simulated scam calls, texts, and emails, so they can practise identifying suspicious requests before facing a real one.

How to Stop Scam Calls Before They Reach You

Reducing the number of scam calls that reach you involves a combination of carrier tools, registry sign-up, and basic call habits. According to the FCC and FTC, no single method blocks every unwanted call, but layering a few approaches meaningfully cuts down exposure.

• Enable carrier-level call blocking: Most major carriers offer built-in or downloadable blocking tools, such as T-Mobile's ScamShield, Verizon's Call Filter, AT&T's Call Protect, and similar services from other providers. The FCC notes these tools use blocking and labeling services that legally stop billions of unwanted and illegal calls each year, and many are free or already included with your plan.
• Register with the National Do Not Call Registry: Adding your number at donotcall.gov is free and takes effect within about 31 days. It's worth setting the right expectation here: the FTC is explicit that the Registry stops sales calls from real companies that follow the law. It does not stop scammers, since illegal callers don't check the list before dialing.
• Use a third-party call-blocking app: For cell phones, the FTC points to a list of blocking and labeling apps maintained by CTIA, the U.S. wireless industry association. These apps flag or block suspected scam numbers before your phone rings.
• Let unknown numbers go to voicemail: Since neither the Registry nor blocking apps catch every spoofed or new number, the simplest fallback is not answering unrecognized calls at all. A legitimate caller will leave a message; a scam call typically won't.
• Report the calls you do receive: Both the FCC and FTC use consumer reports to identify patterns and support carrier blocking systems. Reports can be filed at fcc.gov/complaints or reportfraud.ftc.gov.

Already lost money to a scam like this? Unscammed has helped file over $50 million in reported scams, automating your filing to the FBI and FTC so you don't face it alone.

Frequently Asked Questions

Can voice cloning copy someone's voice from just a phone call, or does it need a recording?

Voice cloning tools work from a pre-recorded audio sample, not a live call. Scammers typically collect clips beforehand from social media, voicemail greetings, or public videos, then generate the clone offline before placing the actual scam call.

Is AI text to speech the same thing as voice cloning?

No. AI text to speech generates generic synthetic speech that isn't tied to any specific person. Voice cloning uses a real person's voice sample to generate speech that sounds like them specifically, which is why it's the version used in impersonation scams.

How is an AI voice changer different from a full voice clone in a scam call?

An AI voice changer alters a scammer's own voice in real time, shifting pitch or tone, while a voice clone replaces it entirely with a synthetic copy of someone else's voice. Voice changers are used more often to disguise identity than to impersonate a specific relative.

Why do AI scams targeting families work even when people know about grandparent scams?

Older scam scripts relied on an unfamiliar voice, which was a natural warning sign. AI scams remove that signal entirely by using a cloned voice the listener recognizes, so the FBI and FTC now recommend verifying identity through a separate method rather than by ear.

Does the National Do Not Call Registry help stop scam calls that use voice cloning?

No. The FTC states clearly that the Registry only stops sales calls from legitimate companies that follow the law. It has no effect on scammers, including those using cloned voices, since illegal callers don't check the list before dialing.

If I can't tell a cloned voice from a real one, what's the most reliable way to stop scam calls from reaching me?

Carrier-level call blocking, third-party blocking apps, and letting unknown numbers go to voicemail are the main tools recommended by the FCC. None of these detect a cloned voice specifically, so they are paired with independent verification once a call does get through.