AI Voice Cloning Statistics 2026: Scams, Deepfakes, Laws and Market Size

The most quoted figures are the FBI's 22,364 AI-related complaints and $893 million in losses for 2025, McAfee's three-second clone and UCL's finding that listeners catch deepfake speech only 73% of the time.

Treat the FBI and FTC figures as floors: they count only what victims reported, and the FBI says many victims do not realize AI was involved. The FTC figure covers all imposter scams, and the McAfee figures come from a 2023 survey in seven countries, so use them as a baseline rather than a current US rate.

Three research firms put the global voice cloning market at $2.40 billion to $2.64 billion in 2025, but their forecasts range from $9.53 billion by 2031 to $31.41 billion by 2035.

The spread comes from each firm's horizon and growth rate, not the base year. Roots Analysis puts 2024 at $1.9 billion, so on its numbers the market grew about $740 million in a year. If you need one figure, quote Mordor Intelligence's $2.40 billion for 2025, because its page is the most recently updated and its base is the lowest of the three. Mordor puts 2026 at $3.02 billion and gives North America 38.7% of 2025 revenue (Mordor Intelligence, 2026).

A few seconds of audio is enough for a rough clone. McAfee researchers produced an 85% voice match from three seconds of audio in 2023 and reached 95% by training on a small number of audio files (McAfee, 2023). The raw material is everywhere: 53% of adults in McAfee's survey share their voice online at least once a week, and a voicemail greeting or one social video holds far more than three seconds.

Consent checks are thin. When Consumer Reports tested voice cloning products from six companies in March 2025, researchers cloned a voice from public audio in four of them, which asked only for a checked box or similar promise that the user had the legal right to clone it (Consumer Reports, 2025). Cost is no barrier either: the deepfake message behind the fake Joe Biden robocall cost $150, according to the FCC (FCC, 2024).

Most people cannot do it reliably. In a UCL study published in PLOS ONE in 2023, 529 listeners spotted deepfake speech only 73% of the time, so about one fake in four fooled them, in English and Mandarin alike (University College London, 2023). Showing listeners examples of deepfakes first improved their scores only slightly.

People know it: 70% of adults told McAfee they were not confident they could tell a clone from the real thing (McAfee, 2023). The FBI warns that a real call and an AI voice clone "can sound nearly identical" (FBI, 2025), so its advice relies on a family code word and a callback, not the listener's ear. Proving a genuine recording is yours is the reverse problem, covered in our guide on how to prove a voice is really you.

About one adult in four has experienced an AI voice scam, by McAfee's count. Its 2023 survey of 7,054 people in seven countries found that 1 in 10 had been targeted personally and 15% said it had happened to someone they know (McAfee, 2023). Of the victims, 77% lost money; more than a third of those who paid lost over $1,000, and 7% lost $5,000 to $15,000. In the United States, 2.2% of the 1,008,597 complaints the FBI IC3 received in 2025 referenced AI (FBI IC3, 2025).

Americans reported $893 million in AI-related losses to the FBI in 2025, about 4.3% of the $20.877 billion in all internet crime losses (FBI IC3, 2025). The FBI counts any complaint that mentions AI, so the total covers fake videos and chatbot scripts too; the figure tied most directly to voice cloning is the more than $5 million lost to "distress" scams, where a cloned voice of a loved one asks for emergency money.

Example: who carries the losses. People 60 and older filed 3,143 of the 22,364 AI-related complaints in 2025, about 14%, but reported $352 million of the $893 million in losses, about 39% (our calculation from FBI IC3, 2025). That is roughly $112,000 per AI-related complaint from someone 60 or older, against about $40,000 across all ages, and nearly three times the $38,500 average loss on all complaints from that age group, whose total internet crime losses rose 59% to $7.748 billion.

The FTC's count is wider because it covers all fraud: 3 million reports and $15.9 billion in losses in 2025, including more than 1 million imposter scam reports and more than $3.5 billion in imposter losses (FTC, 2026). Deloitte's Center for Financial Services projects that generative AI could push US fraud losses to $40 billion by 2027, from $12.3 billion in 2023, or 32% a year (Deloitte, 2024).

The incident that changed US enforcement is the fake Joe Biden robocall: on January 21, 2024, two days before New Hampshire's Democratic presidential primary, 9,581 calls carried an AI imitation of President Biden's voice telling voters to "save your vote for the November election" (FCC, 2024). The $150 message, sent with spoofed caller ID, led to a $6 million forfeiture order against political consultant Steve Kramer, adopted September 26, 2024, under the Truth in Caller ID Act, and the originating carrier, Lingo Telecom, paid a $1 million civil penalty under an August 2024 consent decree (FCC, 2024). Since April 2025 the same tools have been aimed at people with access: scammers send texts and AI voice memos posing as senior US officials to current and former officials and their contacts, building rapport before a malicious link, which drew an FBI public warning on May 15, 2025 (FBI, 2025).

No single federal law covers AI voice cloning as of October 2026. What exists is an FCC robocall ruling, state laws led by Tennessee's ELVIS Act, and the NO FAKES Act, a federal bill that has cleared committee but not the Senate.

The FCC ruling has the most force today because it puts AI voices under an existing federal statute. The NO FAKES Act would go further: the Congressional Research Service summary says it creates a federal intellectual property right in your voice and visual likeness that cannot be transferred during your life, can be licensed and survives your death, with exemptions for news, sports and satire (GovInfo, 2026). It would also preempt state digital replica laws, except claims under laws in force by January 2, 2025, such as the ELVIS Act. Until it passes, protection is a state patchwork, tracked in our guide to digital replica laws by state; our NO FAKES Act explainer covers the federal bill and deepfake laws by country covers the rest of the world.

Americans are worried about AI fakery and do not trust the industry to police itself. In Pew Research Center's 2025 report, 66% of US adults were highly concerned about people getting inaccurate information from AI, and impersonation was among the common worries of both the public and AI experts (Pew Research Center, 2025). Pew found that 59% have not too much or no confidence in US companies to develop and use AI responsibly, and about six in ten worry more that government regulation will not go far enough. Pew surveyed 5,410 US adults in August 2024. That 59% is earned: four of the six cloning products Consumer Reports tested relied on a checkbox for consent.

You cannot stop every clone, so make one useless to a scammer, following FBI guidance where it exists.

1. Set a family code word. The FBI recommends a secret word or phrase that family members use to verify each other (FBI, 2025); keep it off social media.

2. Hang up and call back. If a call or voice note asks for money or codes, call the person back on a number you already have.

3. Never hand over a code. The FBI says never to give a two-factor code to anyone by email, text or messaging app, however familiar the voice.

4. Shrink your public audio. Swap a personal voicemail greeting for the default one and keep long talking videos off public profiles.

5. Put a PIN on your accounts. A verbal password or PIN with your bank and mobile carrier means a caller needs more than your voice to get in.

6. Report every attempt. File at ic3.gov and ReportFraud.ftc.gov, even when you lost nothing.

7. Own the authorized version of your voice. Record it with a service where you decide who hears it and what happens to it later.

Our guide to protecting your voice from AI cloning covers each step in depth, and our guide to whether voice cloning is safe covers cloning your own voice on purpose.

Every scam on this page runs on a voice taken without permission. On Afterlife.ai™, the person whose voice is cloned is the person who signs up, records it and decides who hears it.

1. Start a free build. Sign up at app.afterlife.ai or in the iPhone or Android app: 25 memories to build your Persona, up to 5 Trusted Contacts and Executor Lock™ setup, no card, and your free build never expires.

2. Record your own voice. Creating your voice is free and so is your first listen, so you hear your Persona speak in your own cloned voice before paying.

3. Add your memories and stories. Your Persona answers from the memories you record, in your own words.

4. Name your Trusted Contacts. Choose the people who can talk with your Persona, up to 5 on the free build.

5. Set up Executor Lock™. Name your executor now. At verified death your Persona is locked as a perfect snapshot: nobody can edit, retrain or impersonate you, and your family always meets the Persona you approved.

6. Add a Persona Video. One 30-second recording gives you video messages in your face and your own voice, and your first Persona Video is free.

7. Upgrade only if you need more. Legacy is $14.99 a month and Eternal is $29.99 a month. Afterlife Pro ($59.99 a month or $599 a year) adds Afterlife.ai™ Twin, which answers visitors as you on your website, through an API and in AI assistants over MCP. See pricing.

Two rules keep this consent-first: nobody under 18 can build a Persona, and Afterlife.ai™ never recreates a person who did not consent while alive. Executor Lock™ is the patent-pending governance layer behind that promise, and Idy® has 78 patents pending. For the mechanics, read how to clone your voice; for the legal side, read who owns your voice clone.

• FBI IC3: 2025 Internet Crime Report

• FBI IC3: Senior US officials impersonated in messaging campaign (2025)

• FTC: Joint Economic Committee testimony on fraud (2026)

• FCC: Forfeiture Order, Steve Kramer, FCC 24-104 (2024)

• Help Net Security: McAfee survey on AI voice scams (2023)

• Consumer Reports: Assessment of AI voice cloning products (2025)

• PLOS ONE: Humans cannot reliably detect speech deepfakes (UCL, 2023)

• Pew Research Center: How the US public and AI experts view AI (2025)

• Deloitte: Deepfake banking fraud risk on the rise (2024)

• Mordor Intelligence: Voice cloning market size and share (2026)

• IMARC Group: Voice cloning market report (2026)

• Roots Analysis: Voice cloning market (2026)

• Vanderbilt Law School: Tennessee's ELVIS Act (2024)

• GovInfo: S.4591 bill status and CRS summary, NO FAKES Act of 2026

• Public Citizen: 30 states now have laws to regulate election deepfakes (2026)