AI Deepfakes & Fraud Statistics 2026

Home / Reports / AI Deepfakes & Fraud by the Numbers

Thank you for reading this post, don't forget to subscribe!

Synthetic media has moved from novelty to a line item on the fraud ledger. Deloitte’s Center for Financial Services projects that generative-AI-enabled fraud losses in the United States could reach $40 billion by 2027, up from $12.3 billion in 2023 — a 32% compound annual growth rate (Deloitte, May 2024). The numbers below are drawn only from named primary sources, each cited in-text and listed in full at the end.

The loss curve is steepening

Deloitte’s base case already looks conservative next to law-enforcement data. The FBI’s Internet Crime Complaint Center (IC3) logged more than $16 billion in reported losses across 859,532 complaints in 2024 — a 33% year-over-year jump in dollars lost (FBI IC3 2024 Annual Report). Generative AI does not create a separate fraud category so much as supercharge existing ones: business email compromise, investment scams, and impersonation. Deloitte estimates gen-AI email fraud alone could account for roughly $11.5 billion of US losses by 2027 under an aggressive-adoption scenario.

Incident growth is measured in multiples, not percentages

Identity-verification vendors, which see attack traffic before losses are booked, report the sharpest acceleration. Sumsub recorded a 4x increase in deepfakes detected worldwide from 2023 to 2024 — now 7% of all fraud attempts — followed by a 700% global increase in deepfake fraud between Q1 2024 and Q1 2025 (Sumsub Identity Fraud Report 2024 and 2025). Entrust, whose Onfido unit processes identity checks at scale, found a deepfake attempt occurring every five minutes in 2024, alongside a 244% year-over-year rise in digital document forgeries (Entrust Identity Fraud Report, Nov 2024).

Voice clones and the CEO-fraud playbook

The most expensive single incident of the era remains the engineering firm Arup, whose Hong Kong finance employee wired HK$200 million (about $25.6 million) across 15 transactions after joining a video call populated entirely by deepfake recreations of the CFO and colleagues (CNN, Feb 2024). The delivery channel is shifting to the phone: CrowdStrike observed voice phishing (“vishing”) surge 442% between the first and second halves of 2024 (CrowdStrike 2025 Global Threat Report), a vector tailor-made for cloned-voice social engineering.

The identity-verification front line

Onboarding flows are now a primary battleground. Entrust reports that deepfakes made up 24% of fraudulent attempts to pass motion-based (“liveness”) biometric checks in 2024, and that digital document forgeries surpassed physical counterfeits for the first time, accounting for 57% of all document fraud (Entrust, 2024). Financial services absorbed the brunt, with cryptocurrency platforms drawing nearly double the fraud-attempt rate of any other industry.

Elections: broad reach, uneven impact

Political deepfakes proliferated across the 2024 “year of elections” without producing a single decisive event. Research compiled by Surfshark counts 38 countries that have faced election-related deepfakes since 2021, spanning a combined population of roughly 3.8 billion people. The clearest US case was a January 2024 AI-generated robocall impersonating President Biden’s voice, sent to New Hampshire primary voters (WeLiveSecurity / news of record, 2024).

Detection: humans lose, machines lag

The defensive picture is sobering. In a 2025 iProov study of 2,000 UK and US consumers, only 0.1% correctly identified every real and fake item — even when primed to look for deepfakes (iProov, Feb 2025). A peer-reviewed meta-analysis of 56 studies (86,155 participants) put average human detection accuracy at just 55.5%, barely above a coin flip. Automated tools do better but remain imperfect: the academic Deepfake-Eval-2024 benchmark of real, in-the-wild content found leading detectors reached about 78% accuracy — a meaningful edge over people, but far from a solved problem.

Sources

• Deloitte Center for Financial Services, “Generative AI is expected to magnify the risk of deepfakes and other fraud in banking,” May 2024 — https://www.deloitte.com/us/en/insights/industry/financial-services/deepfake-banking-fraud-risk-on-the-rise.html
• FBI Internet Crime Complaint Center (IC3), “2024 IC3 Annual Report,” 2025 — https://www.ic3.gov/AnnualReport/Reports/2024_IC3Report.pdf
• Sumsub, “Identity Fraud Report 2024” (4x deepfake rise; 7% of attempts), 2024 — https://sumsub.com/fraud-report-2024/
• Sumsub, “Identity Fraud Report 2025-2026” (700% Q1 surge), 2025 — https://sumsub.com/fraud-report-2025/
• Entrust, “Identity Fraud Report 2024” (deepfake every 5 minutes; 244% document forgery; 24% of liveness attempts), Nov 2024 — https://www.entrust.com/company/newsroom/deepfake-attacks-strike-every-five-minutes-amid-244-surge-in-digital-document-forgeries
• CNN, “Finance worker pays out $25 million after video call with deepfake ‘chief financial officer’ (Arup),” Feb 4, 2024 — https://www.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk
• CrowdStrike, “2025 Global Threat Report” (vishing +442% H1 to H2 2024), 2025 — https://www.crowdstrike.com/en-us/press-releases/crowdstrike-releases-2025-global-threat-report/
• iProov, “Deepfake Blindspot: Only 0.1% of People Can Accurately Detect AI-Generated Deepfakes,” Feb 12, 2025 — https://www.iproov.com/press/study-reveals-deepfake-blindspot-detect-ai-generated-content
• Deepfake-Eval-2024, “A Multi-Modal In-the-Wild Benchmark of Deepfakes Circulated in 2024,” arXiv, 2025 — https://arxiv.org/html/2503.02857v1
• Surfshark Research, “38 countries have faced deepfakes in elections,” 2024 — https://surfshark.com/research/chart/election-related-deepfakes
• ESET WeLiveSecurity, “Deepfakes in the global election year of 2024” (Biden robocall), 2024 — https://www.welivesecurity.com/en/cybersecurity/deepfakes-election-year-2024-weapon-mass-deception/