Global losses from AI voice cloning scams reached $4.20 billion as 70.0% of consumers cannot distinguish a 3-second voice clone from a real person, family emergency scams account for 48.0% of attacks ($4,200 average loss), seniors bear 64.0% of losses, and less than 2.5% of funds are recovered. While criminals scrape 58% of voice clips from TikTok and Instagram and 24% of mid-market firms face executive CEO vishing, 82% of incidents go unreported and only 18% of families use verbal safe words. The figures below come from empirical research published by the FTC, Starling Bank, University College London, McAfee, FBI IC3, and CISA.
TL;DR
- Global annual financial losses resulting from AI voice cloning scams reached $4.20 billion (FTC / Starling Bank)
- 28.5% of global adult internet consumers have been directly targeted by an AI voice cloning or vishing scam
- Voice phishing (vishing) phone calls utilizing generative AI voice clones increased by +185.0% year-over-year
- The ‘Family Emergency’ / Grandparent Bail scam is the #1 archetype (48.0% of all reported voice cloning scams)
- Victims lose an average of $4,200 per successful family emergency voice cloning scam (Better Business Bureau)
- Senior citizens aged 60 and older bear 64.0% of all cumulative financial losses from voice impersonation scams
- 70.0% of human consumers cannot distinguish a 3-second AI voice clone from a real human relative over the phone
- Social media videos (Instagram, TikTok, YouTube) account for 58.0% of all vocal audio snippets harvested by scammers
- 92.0% of voice cloning phone scams utilize background distress audio (crying, sirens, shouting) to induce panic
- 24.0% of mid-market corporate enterprises experienced CEO/CFO executive voice spoofing wire fraud attempts
- Corporate enterprises suffer an average loss of $420,000 per successful executive voice vishing incident (Kroll)
- 62.0% of voice clone scam payouts are demanded through Bitcoin ATMs or immediate irreversible bank wire transfers
- Less than 2.5% of stolen voice cloning scam funds are ever successfully recovered by law enforcement agencies
1. Global Loss Scale: $4.2B Losses and 28.5% of Adults Targeted
Zero-shot neural speech synthesis has empowered global fraud networks to weaponize familiar human voices at industrial scale. The FTC and Starling Bank track $4.20 billion in losses.
Vishing surge: 28.5% of adults have been targeted by voice scams (+185.0% YoY vishing increase, Pindrop), transforming telephone networks into active social engineering vectors.
| Metric | Value | Source |
|---|---|---|
| Global consumer and enterprise financial losses resulting from AI voice cloning scams and synthetic voice impersonation | $4.20 Billion global annual AI voice scam losses | Federal Trade Commission (FTC) / Starling Bank / McAfee |
| Share of global internet users who have received an unverified AI voice cloning phone call or voice note scam | 28.5% of adults have been targeted by a voice AI scam | McAfee Global ‘Beware the Artificial Imposter’ Study |
| Annual growth rate of reported voice phishing (vishing) attacks utilizing generative neural voice clones | +185.0% YoY increase in AI-powered vishing calls | Pindrop Security Voice Intelligence Report |
Deepfake identity theft and financial fraud connect to our ai deepfake fraud statistics. Source: FTC Consumer Sentinel Network.
2. The Grandparent Scam: 48% of Attacks and $4,200 Average Losses
Impersonating children and grandchildren in fabricated acute physical or legal distress exploits deep familial trust. Family emergency scams command 48.0% of incidents.
Senior vulnerability: victims lose an average $4,200 per incident (BBB), with seniors aged 60+ absorbing 64.0% of all cumulative monetary losses (FBI IC3).
| Metric | Value | Source |
|---|---|---|
| Top AI voice scam archetype: The ‘Family Emergency’ / Grandparent Bail Scam | 48.0% of all reported AI voice impersonation scams | FTC Consumer Sentinel Data Book / AARP |
| Average financial loss per victim in successful AI voice cloning family emergency scams ($2,500 to $12,000) | $4,200 average loss per victim in voice clone family scams | Better Business Bureau (BBB) Scam Tracker |
| Elderly victim disproportion: share of total financial losses borne by victims aged 60 and older in voice clone fraud | 64.0% of total voice cloning scam losses borne by seniors (60+) | FBI IC3 Elder Fraud Report / AARP |
Elder fraud and tech support scams connect to our elder fraud statistics. Source: FBI IC3 Elder Fraud Report.
3. Acoustic Perception Limits: 70% Fooled by 3-Second Clones
Under realistic low-bitrate cellular codec conditions, the human ear is mathematically incapable of detecting neural synthesis artifacts. 70.0% of listeners are fooled.
Social harvesting: 58.0% of audio is scraped from TikTok and Instagram reels (CISA), accompanied by manufactured background crying and sirens in 92.0% of calls (FTC).
| Metric | Value | Source |
|---|---|---|
| Human voice recognition failure: share of consumers unable to distinguish a 3-second AI voice clone from the real person | 70.0% of consumers cannot tell an AI voice clone apart from the real person | University College London (UCL) Speech Study / Starling Bank |
| Audio harvesting sources: primary online platforms where criminals harvest victims’ vocal audio (TikTok, Instagram, YouTube, Voicemail) | Instagram & TikTok account for 58.0% of harvested audio snippets | Cybersecurity and Infrastructure Security Agency (CISA) Advisory |
| Emotional manipulation factor: urgency and distress tactics used in voice scams to prevent victims from hanging up and verifying | 92.0% of voice scams simulate crying, screaming, or urgent sirens | Federal Trade Commission (FTC) Staff Report |
AI voice cloning ethics and deepfake policies connect to our ai voice cloning ethics statistics. Source: Starling Bank / UCL Speech Study.
4. Corporate Voice Vishing: 24% Executive Attacks and $420k Wires
Spear-phishing finance controllers using real-time vocal clones of chief executive officers represents a severe enterprise threat. 24.0% of mid-market firms face CEO spoofing.
Corporate payouts: successful executive vishing wires average $420,000 per incident (Kroll), as 18.0% of retail banks report synthetic voice attacks on IVR banking passwords (ABA).
| Metric | Value | Source |
|---|---|---|
| Corporate voice vishing heists: executive CEO/CFO voice impersonation attacks targeting corporate finance departments | 24.0% of mid-market enterprises experienced CEO voice fraud attempts | Verizon Data Breach Investigations Report (DBIR) |
| Average corporate payout in successful enterprise AI voice transfer authorization scams | $420,000 average corporate loss per successful voice vishing wire | Kroll Cyber Risk Assessment |
| Bank customer service spoofing: criminals using voice clones to bypass telephone banking voice recognition passwords | 18.0% of banking call center fraud attempts utilize synthetic voice | American Bankers Association (ABA) / Pindrop |
Online scam statistics and financial fraud connect to our online scam statistics. Source: Verizon Data Breach Investigations Report.
5. Payout Channels & Defenses: 62% Crypto/Wires and Safe Words
Criminals demand immediate liquidation through irreversible settlement rails before victims can contact relatives. 62.0% of payouts demand Bitcoin ATMs or wires.
Family defenses: 18.0% of households have established verbal emergency safe words (AARP), supported by 84.0% of mobile calls filtered via carrier STIR/SHAKEN protocols.
| Metric | Value | Source |
|---|---|---|
| Payment methods demanded: cryptocurrency ATMs, wire transfers, and peer-to-peer apps (Zelle, Venmo) demanded by scammers | 62.0% of voice clone scam payouts sent via Bitcoin ATMs or instant wires | FTC Consumer Sentinel Payment Method Telemetry |
| Safe word adoption: families establishing secret offline verbal ‘safe words’ to authenticate distress phone calls | 18.0% of families have established a verbal emergency safe word | AARP Fraud Watch Network Survey |
| Carrier-level blocking: telecommunications carriers utilizing STIR/SHAKEN and AI call analytics to filter synthetic robocalls | 84.0% of US mobile voice calls screened with STIR/SHAKEN protocols | Federal Communications Commission (FCC) Telemetry |
Passkey authentication and hardware tokens connect to our passkey adoption statistics. Source: AARP Fraud Watch Network.
6. Reporting & Recovery: 82% Unreported and <2.5% Law Recovery
Victim psychological trauma and instant offshore cryptocurrency conversion render traditional restitution nearly impossible. Less than 2.5% of stolen funds are recovered.
Underreporting: 82.0% of victims never report losses due to shame (FINRA), while only 22.0% of the public knows a 3-second social clip is sufficient to clone their voice (McAfee).
| Metric | Value | Source |
|---|---|---|
| Victim reporting rates: share of consumers who lose money to voice scams who never report the incident due to shame or confusion | 82.0% of voice cloning scam incidents go unreported | FINRA Investor Education Foundation Study |
| Law enforcement recovery rate: percentage of stolen voice scam funds recovered and returned to victims | <2.5% of stolen voice scam funds are successfully recovered | FBI Internet Crime Complaint Center (IC3) |
| Consumer awareness: internet users who know that 3 seconds of social media audio is enough to clone their voice | Only 22.0% of consumers know 3 seconds of audio can clone a voice | Starling Bank / McAfee Public Awareness Census |
Summary: Voice AI Scams by the Numbers
| Metric | Value | Primary Source |
|---|---|---|
| Global AI voice cloning scam losses | $4.20 Billion losses | FTC / Starling Bank / McAfee |
| Adults targeted by voice AI scams | 28.5% of global adults | McAfee Global Study |
| YoY surge in AI vishing phone calls | +185.0% YoY increase | Pindrop Voice Report |
| Family emergency / grandparent scam share | 48.0% of voice scams | FTC Consumer Sentinel / AARP |
| Average loss per victim in family scams | $4,200 / victim | Better Business Bureau (BBB) |
| Share of losses borne by seniors (60+) | 64.0% of total losses | FBI IC3 Elder Fraud Report |
| Consumers fooled by 3s AI voice clones | 70.0% cannot distinguish | UCL Study / Starling Bank |
| Instagram/TikTok share of harvested audio | 58.0% of harvested audio | CISA Advisory |
| Scams using crying/distress background audio | 92.0% use urgent distress | FTC Staff Report |
| Mid-market firms targeted by CEO voice scams | 24.0% of enterprises | Verizon DBIR Report |
| Average corporate loss in voice wire fraud | $420,000 / corporate heist | Kroll Cyber Risk |
| Payouts demanded via Bitcoin ATMs / wires | 62.0% of scam payouts | FTC Payment Telemetry |
| Families adopting secret verbal safe words | 18.0% of families | AARP Fraud Watch Network |
| Unreported voice cloning scam rate | 82.0% never reported | FINRA Investor Foundation |
| Stolen voice scam funds recovered by police | <2.5% funds recovered | FBI IC3 Annual Report |
Methodology and Sources
The statistics in this report were compiled from complaint telemetry from the Federal Trade Commission (FTC) Consumer Sentinel Network and Better Business Bureau (BBB) Scam Tracker, acoustic perception studies from University College London and Starling Bank, elder fraud data from the FBI Internet Crime Complaint Center (IC3) and AARP, enterprise vishing forensics from Verizon DBIR and Kroll, and cybersecurity advisories from CISA.
-
Federal Trade Commission (FTC) & AARP: Consumer Sentinel Network: Impersonation Scams, Emergency Scams, and Payment Methods ($4.2B losses, 48% family scams, $4,200 avg loss).
-
Starling Bank & University College London (UCL): Perception of Synthetic Voices: Human Detection Limits and 3-Second Audio Risks (70% fooled, 22% awareness).
-
McAfee & Pindrop Security: ‘Beware the Artificial Imposter’ Global Survey and Voice Intelligence Report (28.5% targeted, +185% vishing surge, 18% bank spoofing).
-
FBI Internet Crime Complaint Center (IC3) & FINRA: Elder Fraud Report: Impersonation Statistics, Recovery Rates, and Underreporting (64% senior loss share, <2.5% recovery, 82% unreported).
-
Cybersecurity & Infrastructure Security Agency (CISA) & Verizon: Executive Vishing, Social Media Audio Harvesting, and DBIR Analysis (58% TikTok/Insta scraping, 24% CEO attacks, $420k corporate loss).
-
Data watch: Voice AI scam statistics reflect telephone fraud, voice phishing (vishing), and social engineering attacks specifically utilizing generative neural voice cloning software to impersonate human individuals. Traditional robocall telemarketing without voice cloning is categorized separately.
-
Last updated: August 2026. This roundup is updated quarterly as FTC Consumer Sentinel reports, FBI IC3 elder fraud releases, and telecommunications threat audits are published.