Deepfake Fraud Statistics 2026: Losses, Attacks, and Detection

Key Findings

• Investment scams accounted for 70.7% of the losses in AI-related complaints to the FBI in 2025.
• People 60 and older filed 14.1% of AI-related complaints but reported 39.5% of the losses, an average of $112,153 per complaint.
• 62% of organizations in a Gartner survey of 302 cybersecurity leaders had experienced a deepfake attack in the prior 12 months.
• About 25% of US enterprises in Pindrop's 2026 survey that had experienced or suspected a deepfake attack lost $1 million or more.
• Across 56 studies, people correctly identified deepfakes 55.54% of the time, a rate not significantly better than chance.

How Much AI-Related Fraud Cost Americans in 2025

The FBI's 2025 Internet Crime Report is the first to count AI-related complaints as their own category, so we broke those complaints down by the type of crime each victim reported to find which scams cost Americans the most. The FBI tags a complaint as AI-related when the information reported contains a reference to AI, so the count depends on victims knowing or suspecting that AI was used.² IC3 counted 22,364 of these complaints with $893,346,472 in adjusted losses. That is 2.2% of all complaints and 4.3% of all reported losses.²

Investment scams made up $632,041,188 of the total. The FBI says investment clubs use AI-generated videos and voices of celebrities, CEOs, and other trusted figures in fake endorsements on social media and in video calls.²

Source: FBI IC3, 2025 Internet Crime Report.²

Key Findings

• Only 135 AI-related complaints involved business email compromise, but they averaged $224,123, the highest average of any AI-related crime type.
• Investment scams averaged $145,097 per complaint, about 11 times the $12,807 average for phishing or spoofing.
• AI-related complaints that also involved cryptocurrency reported $658,714,247, or 73.7% of AI-related losses.²

The FBI notes that all investment scam losses in 2025 exceeded $8 billion, demonstrating that many victims do not realize the extent AI may be involved in scams.² AI-related complaints made up 7.3% of the $8,648,617,756 in investment losses. In job scams, IC3 describes online interviews where the candidate's lip movements do not match the audio, and says the goal generally appears to be access to private computer networks. Victims still reported $12.6 million in AI-involved employment scam losses.²

How AI-Related Fraud Hit Americans 60 and Older

Warnings about voice-cloning and fake-video scams are often aimed at older relatives, so we separated out the AI-related complaints that people 60 and older filed with the FBI and compared their losses with everyone else's. We also added iProov's 2025 consumer test, which asked people of different ages whether they had heard of deepfakes. Complainants 60 and older filed 3,143 AI-related complaints in 2025 and reported $352,496,231 in losses.² That is 14.1% of AI-related complaints and 39.5% of AI-related losses, for an average of $112,153 per complaint, 2.8 times the average for AI-related complaints at all ages. Across every crime type, people 60 and older filed 201,266 complaints and reported $7.7 billion in losses, an average of $38,500.²

Key Findings

• AI-related complaints averaged about 1.9 times the loss of all complaints, and those from people 60 and older averaged about 5.4 times as much.
• AI-related complaints accounted for about 4.5% of the $7.7 billion that people 60 and older reported losing.
• Victims reported more than $5 million in 2025 losses to distress scams, in which cloned voices mimic a relative or friend in an emergency.²

Awareness is lowest among older adults. In iProov's 2025 test of 2,000 UK and US consumers, 22% had never heard of deepfakes, including 30% of people aged 55 to 64 and 39% of those 65 and older.²⁰

How Many Organizations Report Deepfake Attacks

Before deciding how much to spend on deepfake defenses, a company needs some sense of how often attackers try them. Three surveys of security and finance leaders, from Gartner, Regula, and Deloitte, asked whether their organizations had been hit. In Gartner's survey of 302 cybersecurity leaders in North America, EMEA, and Asia/Pacific, conducted from March to May 2025, 62% said their organization had experienced a deepfake attack in the prior 12 months that used social engineering or exploited automated processes.³⁴ 43% reported at least one deepfake audio call incident, and 37% had experienced deepfakes in video calls.⁴

Regula's 2025 survey found deepfake fraud at 33% of organizations, close to the rates for document fraud and social engineering scams, at 30% each.⁵ In a Deloitte poll of more than 1,100 executives in May 2024, 15.1% said their organization had one deepfake financial fraud incident in the past year and 10.8% had more than one.⁶

Key Findings

• The three surveys asked different questions of different groups, so their rates are not directly comparable. In each, deepfakes reached at least a quarter of organizations.
• 40% of companies in Regula's survey with $1 million or more in fraud losses said they had been hit by deepfakes.⁵
• 51.6% of Deloitte's polled executives expected the number and size of deepfake attacks on their financial and accounting data to grow over the next 12 months.⁶

What Deepfake Attacks Cost Businesses

A security budget is easier to set when there is a dollar figure attached to the risk, and several studies now put one on deepfake attacks. The figures here come from Pindrop's and Regula's surveys, IBM's 2026 breach cost study, Resemble AI's tracking of incidents in the news, and CNN's reporting on the largest confirmed corporate case. Pindrop's Deepfake Readiness Index, a Wakefield Research survey of 250 US security leaders at organizations with at least 1,000 employees conducted in June 2026, found that almost half of the organizations that had experienced or suspected a deepfake attack reported total costs of $500,000 or more.⁷ About 25% crossed $1 million, counting direct losses, remediation, and staff time, and 49% faced follow-on cyberattacks such as ransomware.⁷

Sources: Pindrop, Regula, IBM, and CNN.^7101112

Key Findings

• In Regula's 2024 survey, 56% of businesses said they were very confident they could detect deepfakes, yet only 6% had avoided financial losses from them.¹⁰
• More than one in four organizations in IBM's 2026 study that had a malicious attack said it was AI-driven, up 56% from the year before, and deepfake impersonation made up 45% of those AI-driven attacks.¹¹
• Resemble AI confirmed 821 deepfake incidents in news coverage in the first half of 2026, with $6.95 million in confirmed direct losses. Only 15 were corporate fraud, and 14 of those were covered by a single source.¹³

Arup, the British engineering firm, confirmed in May 2024 that fake voices and images were used in the Hong Kong case, in which, according to Hong Kong police, a finance worker sent the money after a video call with people he believed were the chief financial officer and other staff, all of them deepfake re-creations.¹²

How Fast Deepfake Attempts Are Growing

Defenses sized for last year's attack volume can fall behind quickly, which is why the growth rate matters as much as the count. Call center and identity verification firms Pindrop, Entrust, and Sumsub each publish how often they see deepfakes in the systems they protect, and FinCEN's 2024 alert adds what banks have been reporting. Across Pindrop's customer base, AI attacks rose 1,210% by December 2025, compared with Q4 2024.²²⁸ By Q1 2026 the increase was 1,390%, and by the end of June 2026 it had reached 1,680%. Pindrop says AI attacks are growing 8 times faster than traditional attack types.⁸ In Q1 2026, 8.7% of contact center fraud across its customers was AI-backed, up from under 1% a year earlier, with an average of 287 AI attacks a day.⁹

Sources: Entrust, Sumsub, and Pindrop.^14151617

Key Findings

• Sumsub's Q1 2025 deepfake fraud increases were larger outside the US, led by Canada at 3,400% and Hong Kong at 1,900%.¹⁵
• Digital forgeries made up 35% of the document fraud Entrust saw in 2025, up from a 29% average from 2022 to 2024.¹⁴
• FinCEN told financial institutions in November 2024 that, beginning in 2023, it had seen more suspicious activity reports describing suspected deepfake media, often in altered or fabricated identity documents.¹⁸

How Accurately People Spot Deepfakes

Whether a company can lean on its staff to catch a fake caller depends on how good people are at spotting deepfakes, and researchers have tested that directly. We drew on a 2024 meta-analysis by Diel and colleagues and iProov's 2025 consumer test, then on a 2024 Deloitte poll and Pindrop's 2026 survey for the tools organizations use instead. Diel and colleagues pooled 56 papers with 86,155 participants and found that people correctly identified deepfakes 55.54% of the time, a rate not significantly above chance because its 95% confidence interval crossed 50%.¹⁹ Accuracy was highest for audio, at 62.08%, though from only eight effects with a wide confidence interval, and lowest for text, at 52.00%. Feedback training, AI support, and similar strategies raised accuracy on deepfakes to 65.14%.¹⁹

Key Findings

• In iProov's test of 2,000 UK and US consumers who were told to look for deepfakes, 0.1% correctly identified every real and fake image and video.²⁰
• iProov's participants were 36% less likely to correctly identify a synthetic video than a synthetic image.²⁰
• Detection technology was the primary defense for 6.8% of Deloitte's polled executives at organizations with no deepfake incidents and 15% of those at organizations with more than one.⁶

Few companies have bought tools built for the job. Only 10% of the US security leaders in Pindrop's 2026 survey said their organization had invested in purpose-built deepfake detection, 37% rely on legacy tools, and 93% report they are at least somewhat concerned about deepfake attacks.⁸

How Much Generative AI Fraud Could Cost by 2027

Deloitte and Gartner have both published forecasts for AI-enabled fraud through 2027, and they are worth reading together because the budgets and controls set this year will have to hold up against those losses. 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, a compound annual growth rate of 32%.²¹ It estimates that generative AI email fraud losses alone could reach about $11.5 billion by 2027 in an aggressive adoption scenario.²¹

Gartner predicts that by 2027 AI agents will halve the time needed to exploit account takeovers, with steps such as deepfake voice calls automated, and that 50% of enterprises will be investing in disinformation security, up from less than 5%.⁴ In a 2024 Gartner survey of 456 CEOs and senior executives, 62% expected deepfakes to create at least some operating costs and complications for their organizations within three years.⁴

What the Deepfake Fraud Figures Mean

What these figures mean in practice depends on where deepfakes reach people: a phone call from a relative, a video meeting with a manager, or a job interview over a webcam. The FBI's count of AI-related losses is the low end of the problem, since it includes only complaints in which victims mentioned AI, and the bureau says many investment scam victims do not realize AI was involved.

For a business, a familiar voice or face on a call no longer proves who is on the other end. The controls that hold up are the ones that do not depend on anyone recognizing a fake, such as calling back on a number kept on file or requiring a second approval before money moves.

So if you approve payments, hire remote staff, or verify customers online, treat a familiar voice or face on a call as unconfirmed until you check it through a second channel you trust.

Sources

1. CyberReviewed, Deepfake Fraud Study: analysis of AI-related complaint counts and adjusted losses by crime type and age group from the FBI IC3 2025 Internet Crime Report, which covers 1,008,597 complaints in total, with our calculations of average loss per complaint and shares of AI-related losses, combined with deepfake attack and cost surveys from Gartner, Pindrop, Deloitte, and Regula, attack data from Pindrop, Entrust, Sumsub, and Resemble AI, IBM's 2026 breach cost study, FinCEN's 2024 deepfake alert, and a meta-analysis of human deepfake detection studies, conducted October 2026.
2. FBI Internet Crime Complaint Center, 2025 Internet Crime Report
3. Gartner, Gartner Survey Reveals GenAI Attacks Are on the Rise, September 22, 2025
4. Gartner, Why CIOs Can't Ignore the Rising Tide of Deepfake Attacks, September 2, 2025
5. Regula, survey on deepfakes and traditional fraud, September 2025
6. Deloitte, Half of Executives Expect More Deepfake Attacks on Financial and Accounting Data in Year Ahead, September 17, 2024
7. Pindrop, The Deepfake Readiness Index: Cost of a Deepfake Attack, 2026
8. Pindrop, The Deepfake Readiness Index: Deepfake Detection Tools, 2026
9. Pindrop, Contact Center Attacks, Q1 2026 data
10. Regula, survey on the cost of deepfake fraud, 2024
11. IBM, Cost of a Data Breach Report 2026
12. CNN, Arup revealed as victim of $25 million deepfake scam involving Hong Kong employee, May 16, 2024
13. Resemble AI, H1 2026 Deepfake Threat Report
14. Entrust, 2026 Identity Fraud Report release, November 18, 2025
15. Sumsub, Q1 2025 identity fraud trends, June 12, 2025
16. Sumsub, Identity Fraud Report 2025 release
17. Pindrop, 2025 Voice Intelligence and Security Report release
18. FinCEN, Alert on Fraud Schemes Involving Deepfake Media Targeting Financial Institutions, FIN-2024-Alert004, November 13, 2024
19. Diel et al., Human performance in detecting deepfakes: A systematic review and meta-analysis, Computers in Human Behavior Reports, 2024
20. iProov, Study Reveals Deepfake Blindspot, February 2025
21. Deloitte Insights, Deepfake banking and AI fraud risk
22. Pindrop, AI Fraud Surged 1210%, 2026

Methodology

FBI figures come from the descriptor tables in IC3's 2025 Internet Crime Report, which counts a complaint as AI-related when it references artificial intelligence. Averages, shares, and multiples are calculated from published figures and rounded once. Deloitte's 25.9% adds the shares reporting one and more than one incident. Survey and vendor figures cover different respondents, regions, and periods. The cost figures measure different things, from survey averages to a single case, so they are not directly comparable. Money is in US dollars, and the Arup figure uses CNN's conversion of HK$200 million.