Deepfake Statistics 2026: 30+ Sourced Facts on Fraud, Detection, and Growth

---

Every figure on this page was checked against its primary source before publication, and every stat links to that source with its publication year. If we could not trace a number to the organization that originally published it, we left it out. That is the standard most deepfake statistics roundups skip, and it is the reason you can cite this page with confidence.

Key deepfake statistics for 2026: deepfake files grew from roughly 500,000 in 2023 to a projected 8 million in 2025, a deepfake fraud attempt occurred every five minutes in 2024, and only 0.1 percent of people can reliably tell AI-generated media from real. All figures below link to primary sources.

Editor's Picks: The 10 Deepfake Statistics Everyone Cites

1. $25.6 million stolen from engineering firm Arup in a single deepfake video call scam in Hong Kong, per CNN reporting (2024).
2. A deepfake fraud attempt occurred every 5 minutes in 2024, per the Entrust 2025 Identity Fraud Report (2024).
3. Only 0.1% of people correctly identified all deepfake and real content shown to them in an iProov study of 2,000 consumers (2025).
4. Humans caught only 24.5% of high-quality deepfake videos in the Korshunov and Marcel study at the Idiap Research Institute (2020).

1. Deepfake fraud attempts in contact centers surged more than 1,300% in 2024, from about one per month to seven per day, per Pindrop's 2025 Voice Intelligence and Security Report (2025).
2. Over $200 million was lost to deepfake-enabled fraud in Q1 2025 alone across 163 documented incidents, per the Resemble AI Q1 2025 Deepfake Incident Report (2025).
3. Nearly $900 million in losses across 22,000+ complaints were tied to AI-assisted cybercrime in the FBI's 2025 IC3 Annual Report, the first year IC3 tracked AI as a crime descriptor (2025 data, published 2026).
4. $40 billion: projected US fraud losses enabled by generative AI by 2027, up from $12.3 billion in 2023, a 32% compound annual growth rate, per the Deloitte Center for Financial Services (2024).
5. 3 seconds of audio is enough to clone a voice with an 85% match, per McAfee's Beware the Artificial Impostor report (2023).
6. Roughly 500,000 deepfake files online in 2023 grew to a projected 8 million in 2025, a trajectory first published by DeepMedia and since cited in Europol reporting.

Deepfake Volume and Growth Statistics

1. Deepfake files online grew about 16x in two years, from roughly 500,000 in 2023 to a projected 8 million in 2025 (DeepMedia projection; see Editor's Pick 10 for sourcing). No precise global count exists, and this is the best-documented trajectory.
2. Deepfake fraud in the United States surged 1,100% in Q1 2025, per the Sumsub Identity Fraud Report 2025-2026 (2025).
3. Deepfakes now account for 7% of all fraudulent activity globally detected on Sumsub's platform, per the same Sumsub report (2025).
4. Sophisticated fraud attacks combining multiple techniques rose 180% globally in 2025, per Sumsub (2025).
5. Digital document forgeries rose 244% year over year and now make up 57% of all document fraud, per the Entrust 2025 Identity Fraud Report (2024).
6. 49% of businesses worldwide reported both audio and video deepfake fraud in 2024, up from 37% (audio) and 29% (video) in 2022, per the Regula Deepfake Trends 2024 survey (2024).

Deepfake Fraud and Scam Statistics

1. US cybercrime losses reached $20.9 billion in 2025, up 26% year over year across more than 1 million complaints, per the FBI 2025 IC3 Annual Report (2026 publication).
2. AI-assisted cybercrime drew 22,000+ complaints and nearly $900 million in losses in 2025, the first year the FBI tracked AI as a formal crime descriptor, per IC3 (2026 publication).
3. Americans aged 60 and over lost $7.7 billion to online fraud in 2025, an increase of about 60% from 2024, per IC3 (2026 publication).
4. Investment fraud cost $8.6 billion and cryptocurrency-related fraud $11.3 billion in 2025, the categories where deepfake celebrity endorsements concentrate, per IC3 (2026 publication).
5. Consumers reported $15.9 billion in fraud losses to the FTC in 2025, a record, up from $12.5 billion in 2024 and up nearly 430% since 2020. VERIFY-URL: FTC press release, March 2026; 2024 baseline: [FTC March 2025 release]
6. Impostor scams were the most reported fraud type of 2025 with over $3.5 billion in losses, per FTC 2025 data.
7. Reports of older adults losing $10,000+ to impersonation scams more than quadrupled from 2020 to 2024, and combined losses above $100,000 grew eight-fold from $55 million to $445 million, per the FTC impersonation data spotlight (2025).
8. The Arup deepfake scam moved $25.6 million through 15 wire transfers in one day after an employee joined a video call where every other participant was AI-generated, per CNN (2024).
9. Generative AI could drive US fraud losses to $40 billion by 2027 (conservative scenario: $22 billion), per Deloitte (2024).
10. Public figures were impersonated in 47% of documented deepfake incidents in Q1 2025, while private citizens were 34% of victims, per Resemble AI (2025).

Deepfake Detection Statistics: Humans vs Machines

1. People correctly identified only 24.5% of high-quality deepfake videos in the Korshunov and Marcel human perception study at Idiap (2020). Good fakes do not look fake.
2. Only 0.1% of 2,000 primed participants told all real and fake media apart in iProov's deepfake detection study (2025).
3. Participants were 36% less likely to spot a fake video than a fake image, and 22% of consumers had never heard of deepfakes, per the same iProov study (2025).
4. Biometric injection attacks surged 1,151% year over year in the second half of 2025, per the iProov Threat Intelligence Report 2026 (2026).
5. DeepfakeDetector.ai detects AI-generated video, image, and audio with high accuracy, our published platform benchmark (2026). Machines are not perfect, but they outperform unaided human judgment by a wide margin.
6. First-party platform data: in Q2 2026, 41% of files analyzed on DeepfakeDetector.ai came back likely AI-generated, up from 36% in Q1 2026 and 29% in Q3 2025. It is the one figure on this page no competitor can publish.

Voice Cloning and Audio Deepfake Statistics

1. Three seconds of audio produced an 85% voice match, and a few more clips pushed it to high accuracy, per McAfee's Beware the Artificial Impostor report (2023).
2. 70% of people were not confident they could tell a cloned voice from the real thing, in McAfee's survey of 7,054 adults across seven countries (2023, same report).
3. 1 in 4 adults had experienced an AI voice scam or knew someone who had, and 77% of those victims lost money, per McAfee (2023).
4. Synthetic voice attacks rose 475% at insurance companies and 149% at banks in 2024, from Pindrop's analysis of 1.2 billion customer calls, per the Pindrop 2025 report (2025).
5. Pindrop projected deepfake-related fraud to grow another 162% in 2025, with contact centers facing an estimated $44.5 billion in fraud exposure, per the same 2025 report (2025).
6. Voice cloning now needs only 3 to 5 seconds of sample audio to produce a convincing clone, per Resemble AI's Q1 2025 incident analysis (2025).
7. The FTC ran a Voice Cloning Challenge to surface detection and prevention ideas, announced in its November 2023 consumer alert, a regulatory signal of how seriously the agency treats AI voice fraud (2023).

For documented case files behind these numbers, see our guide to voice cloning scams.

Deepfake Statistics by Industry: Banking, KYC, and Media

1. Deepfakes made up 24% of fraudulent attempts against motion-based biometric verification (the selfie checks banks use in onboarding), per the Entrust 2025 Identity Fraud Report (2024).
2. Synthetic identity document fraud rose more than 300% in the US, with AI tools like ChatGPT and Gemini implicated in 2% of all falsified documents processed globally, per Sumsub (2025).
3. Audio deepfakes dominate in financial services (51% of firms hit), while video deepfakes lead in technology (57%), per the Regula Deepfake Trends 2024 survey (2024).
4. Non-consensual explicit content was the largest deepfake misuse category at 32% of incidents, ahead of financial fraud at 23%, per Resemble AI (Q1 2025).

For the verification side of these numbers, see our guide to KYC deepfake detection.

Deepfake Detection Market Size and Forecast

1. The global deepfake detection market was valued at $114.3 million in 2024 and is forecast to reach about $5.6 billion by 2034, a 47.6% CAGR, per Market.us (2025).
2. Forecasts diverge widely by market definition: Market Research Future projects a 16.33% CAGR for AI deepfake detector tools through 2035, while broader "deepfake AI" market forecasts run above 41% CAGR (Polaris Market Research, 2024). Methodology caveat: scope, base year, and segment definitions differ across firms, so cite the firm and definition, never a bare number.

Note for citers: a widely circulated "$5.5 billion in 2023 to $15.7 billion in 2026" market figure appears in many roundups without a named research firm. We could not trace it to a primary source, so it does not appear in our numbered list. That is exactly the laundering problem this page exists to fix.

How We Verify These Deepfake Statistics (and How to Cite Us)

Methodology. Every statistic above was traced to the organization that originally published it: government data (FBI IC3, FTC), named vendor research with disclosed methodology (Entrust, Sumsub, Pindrop, Regula, McAfee, iProov, Resemble AI), peer-reviewed or preprint academic work, and named market research firms. We exclude figures that circulate only in secondary roundups, and we link the specific report or release, not a homepage. Statistics are re-verified quarterly; the last-verified date sits at the top of this page. Items still awaiting a final URL check are flagged until confirmed.

How to cite us. You are welcome to cite any statistic or embed any chart on this page with attribution:

Suggested citation: DeepfakeDetector.ai, "Deepfake Statistics 2026: 30+ Sourced Facts on Fraud, Detection, and Growth," updated June 12, 2026, https://deepfakedetector.ai/blog/deepfake-statistics

All five charts are original and free to republish with a link back to this page as the source. Journalists on deadline: if you need a number checked or a custom cut of our first-party detection data, email us and we will verify it against the primary source for you.

CTA: Writing about deepfakes? Cite any stat with a link back, and test our detector free, 50 detections per month. Create a free account

FAQ

How many deepfakes are on the internet? No exact count exists. The best-documented projection puts deepfake files at roughly 8 million in 2025, up from about 500,000 in 2023, a trajectory published by DeepMedia and cited in Europol reporting.

How much money has been lost to deepfake fraud? There is no single global total. The clearest sourced markers: nearly $900 million in AI-assisted cybercrime complaints in the FBI's 2025 IC3 report, over $200 million in documented deepfake fraud in Q1 2025 alone per Resemble AI, and $25.6 million in the single Arup case in 2024.

What percentage of people can spot a deepfake? Very few. Humans correctly identified only 24.5% of high-quality deepfake videos in the 2020 Korshunov and Marcel study, and only 0.1% of participants in iProov's 2025 study identified all real and fake media correctly.

How fast is deepfake fraud growing? Pindrop measured a more than 1,300% surge in contact-center deepfake fraud in 2024, and Sumsub recorded a 1,100% jump in US deepfake fraud in Q1 2025. Both firms publish annual methodology, linked above.

How often is this page updated? Quarterly. The last-verified date at the top of the page changes with every review, and stats that fail re-verification are removed or corrected.

Conclusion: Use the Numbers, Check the Sources

These deepfake statistics tell one story: synthetic media fraud is growing faster than human detection ability, and the gap is the whole problem. Use any figure here with its source and year, link back so your readers can verify, and check this page each quarter for refreshed data. And if a suspicious file lands on your desk while you are writing, run it through our deepfake detection platform first.

CTA: Test the detector behind our first-party data, free: 50 detections per month. Get started free

Related reading