On this page
How to read these numbers
There is no official census of deepfakes. The figures come from three kinds of sources, which should not be mixed up:
- public estimates (governments, regulators such as Ofcom);
- data from security and identity verification companies (Sumsub, iProov, Entrust), which measure what they see across their own customers, not the whole internet;
- surveys and studies (Ofcom, Thorn, universities), which measure exposure and perception.
Each figure below comes with its exact scope, date and link. We do not round them up. Growth percentages often start from a low base: a 700% rise means "eight times as many", not "700 times as many".
Key figures at a glance
| Figure | What exactly it measures | Source and date |
|---|---|---|
| 8 million | Deepfakes estimated to have been shared online in 2025, up from 500,000 in 2023 | UK Home Office, February 5, 2026 |
| +700% | Deepfake identity fraud attempts, worldwide, Q1 2025 vs Q1 2024 | Sumsub internal data, via GPO Magazine, June 19, 2025 |
| +900% | Same measure, Europe, Q1 2025 vs Q1 2024 | Sumsub, via GPO Magazine, June 19, 2025 |
| 98% | Share of deepfake videos online that are pornographic (2023) | Security Hero, "2023 State of Deepfakes" |
| 99% | Share of people targeted in deepfake pornography who are women (2023) | Security Hero, "2023 State of Deepfakes" |
| 24 million | Unique visitors in September 2023 to 34 AI "undressing" sites | Graphika, "A Revealing Picture", December 8, 2023 |
| 3 seconds | Audio needed to clone a voice with an 85% match | McAfee, "Artificial Imposters" |
| 0.1% | Participants who correctly identified all real and fake content (2,000 people, UK and US) | iProov, February 12, 2025 |
| 43% | UK people aged 16+ who saw at least one deepfake online in the previous six months | Ofcom, July 23, 2024 |
Volume: how many deepfakes are out there?
| Figure | Scope | Source |
|---|---|---|
| 500,000 deepfakes shared in 2023 | Estimate, worldwide | UK Home Office, "Government leads global fight against deepfake threats", February 5, 2026 |
| 8,000,000 deepfakes shared in 2025 | Estimate, worldwide ("In 2025 alone, 8,000,000 deepfakes are estimated to have been shared, up from 500,000 in 2023") | UK Home Office, February 5, 2026. The same figure appeared as a projection in a UK government case study published February 5, 2025. |
| 95,820 deepfake videos online in 2023, up 550% on 2019 | Videos found across 85 channels and 100+ websites | Security Hero, "2023 State of Deepfakes" |
| 2,031 verified incidents in Q3 2025 | Publicly reported incidents tracked by the company | Resemble AI, "Q3 2025 Deepfake Report", October 2025 |
Going from 500,000 to 8 million is a sixteen-fold increase in two years (our calculation from the Home Office figures).
Fraud and identity theft
| Figure | Exact scope | Source |
|---|---|---|
| +700% | Deepfake identity fraud attempts, worldwide, Q1 2025 vs Q1 2024 | Sumsub internal data, via GPO Magazine, June 19, 2025 |
| +900% | Same measure, Europe | Same |
| +1,100% | Same measure, Asia-Pacific | Same |
| +700% | Deepfake incidents in fintech in 2023 vs 2022 | Sumsub data cited by the Wall Street Journal (April 3, 2024) and Deloitte (May 29, 2024) |
| 10x | Deepfakes detected across all industries, worldwide, 2022 to 2023; +1,740% in North America, +780% in Europe (incl. UK) | Sumsub Identity Fraud Report 2023, November 28, 2023 |
| +704% | Face swap attacks on remote identity verification, H1 to H2 2023 | iProov Threat Intelligence Report, February 7, 2024 |
| One every five minutes | Rate of deepfake attack attempts in 2024 | Entrust / Onfido 2025 Identity Fraud Report, November 18, 2024 |
| More than $200 million | Documented deepfake-related financial losses in Q1 2025 | Resemble AI Q1 2025 report, April 16, 2025 |
| $40 billion by 2027 | Projected generative-AI-enabled fraud losses in the US, up from $12.3 billion in 2023 (32% compound annual growth) | Deloitte Center for Financial Services, May 29, 2024 |
| About $25 million | Paid out by an Arup employee in Hong Kong after a video call with deepfaked executives (2024) | AI Incident Database, Incident 634 |
Sexual deepfakes and abuse of women
| Figure | Scope | Source |
|---|---|---|
| 98% | Share of deepfake videos online that are pornographic (2023) | Security Hero |
| 99% | Share of women among people targeted in deepfake pornography | Security Hero |
| Under 25 minutes, $0 | Time and cost to make a 60-second fake porn video from one clear photo | Security Hero |
| 24 million | Unique visitors in September 2023 to 34 AI "undressing" services | Graphika, December 8, 2023 |
| More than 2,000% | Rise in referral link spam for these services on Reddit and X since the start of 2023 | Graphika, December 8, 2023 |
| About 3 million | Sexualized images produced by Grok's image tool on X in 11 days (December 29, 2025 – January 8, 2026), including about 23,000 that appeared to depict children | Center for Countering Digital Hate, reported by ITV News, January 23, 2026 |
| 4.5 to 1 | Ratio of women to men among people targeted by deepfakes in Q3 2025 | Resemble AI, Q3 2025 |
| 14% | Of UK adults who had seen a deepfake, share who had seen a sexual one | Ofcom, July 2024 |
Children and teens
| Figure | Scope | Source |
|---|---|---|
| 3,443 | AI-generated child sexual abuse videos identified in 2025, a more than 260-fold increase on 2024 | Internet Watch Foundation, 2025 annual report |
| 1 in 8 | Teens (13-17) who know someone targeted by AI-generated deepfake nudes | Thorn, March 3, 2025 (survey of 1,200 young people aged 13-20) |
| 50% | UK children aged 8-15 who saw at least one deepfake in the previous six months | Ofcom, July 2024 |
| 16.3% | Share of Q3 2025 deepfake incidents involving minors (331 cases) | Resemble AI, Q3 2025 |
If a minor is involved: use Take It Down (NCMEC) in the US or Report Remove (Childline and the IWF) in the UK. Our page for under-18s.
Exposure and ability to spot them
| Figure | Scope | Source |
|---|---|---|
| 43% | UK people aged 16+ who saw at least one deepfake in the previous six months | Ofcom polling, June 2024, published July 23, 2024 |
| 9% | UK people aged 16+ confident they can identify a deepfake | Ofcom, July 2024 |
| 0.1% | Participants who correctly classified all real and fake content (2,000 people, UK and US) | iProov, February 12, 2025 |
| 36% | How much less likely people were to spot a fake video than a fake image | iProov, February 12, 2025 |
| Over 60% | Participants who were confident in their detection skills despite poor results | iProov, February 12, 2025 |
| 73% | Rate at which human listeners detected synthetic speech (529 participants) | University College London, August 2, 2023 |
| 3 seconds | Audio needed for an 85% voice match clone | McAfee |
| 1 in 4 | People who experienced an AI voice cloning scam or knew someone who had (7,000 people, 9 countries incl. US and UK); 77% of victims lost money | McAfee |
Legal milestones
| Date | What changed | Where |
|---|---|---|
| May 21, 2024 | France's SREN law: sharing AI-generated content of a person without consent becomes a criminal offense, with a specific offense for sexual deepfakes | France |
| February 6, 2026 | Creating an intimate deepfake of an adult without consent becomes an offense | England and Wales |
| May 19, 2026 | TAKE IT DOWN Act: platforms must remove non-consensual intimate images, including AI forgeries, within 48 hours of a valid request | United States |
| August 2, 2026 | AI Act Article 50: deepfakes must be disclosed as AI-generated or manipulated | European Union |
General information, not legal advice. Deepfake laws explained.
What these numbers do not tell you
- How many people are victims. No government publishes a reliable count of deepfake victims yet.
- Deepfakes nobody reports. Many victims never find the content or never report it. Every count is a floor, not a ceiling.
- Company data is not neutral. Sumsub, iProov and Resemble AI sell detection products. Their figures are useful but describe their own activity.
Behind every number, a person
Every deepfake counted here targets someone. If there is a deepfake of you online, we handle the reports to platforms and Google, the follow-ups and a search for copies every week, for a fixed price. Get a deepfake of you removed.
See pricingFAQ
How many deepfakes are there in 2026?
There is no count for 2026 yet. The latest official estimate is the UK Home Office's: 8 million deepfakes shared in 2025, up from 500,000 in 2023.
Have deepfakes really increased by 700%?
Yes for one specific measure: according to Sumsub, deepfake identity fraud attempts rose 700% worldwide between Q1 2024 and Q1 2025. Sumsub data also showed a 700% rise in deepfake incidents in fintech in 2023. Neither figure measures all deepfakes online.
What share of deepfakes is pornographic?
98% of deepfake videos online in 2023, according to Security Hero, and 99% of the people targeted were women.
Can people tell deepfakes apart from real content?
Poorly. In iProov's 2025 study only 0.1% of participants got everything right, and only 9% of UK adults told Ofcom they felt confident spotting one. How to spot a deepfake.
How much does deep fake fraud cost?
Resemble AI documented more than $200 million in losses in Q1 2025 alone. Deloitte projects that generative-AI-enabled fraud could reach $40 billion in the US by 2027.
Can I use these figures?
Yes, please cite the original source listed for each figure. You can also cite this page (SOS Deepfake, "Deepfake statistics 2026", updated October 5, 2026).
Sources
- UK Home Office, "Government leads global fight against deepfake threats", February 5, 2026
- UK Government case study, "Innovating to detect deepfakes and protect the public", February 5, 2025
- GPO Magazine, "Usurpation d'identité en France : les deepfakes en hausse de 700 %" (Sumsub data, in French), June 19, 2025
- Sumsub, "Global Deepfake Incidents Surge Tenfold from 2022 to 2023", November 28, 2023
- Deloitte Center for Financial Services, "Generative AI is expected to magnify the risk of deepfakes and other fraud in banking", May 29, 2024
- iProov Threat Intelligence Report 2024, February 7, 2024
- iProov, "Deepfake Blindspot" study, February 12, 2025
- Entrust / Onfido 2025 Identity Fraud Report, November 18, 2024
- Resemble AI, Q1 2025 AI Deepfake Security Report, April 16, 2025
- Resemble AI, Q3 2025 Deepfake Report, October 2025
- Security Hero, "2023 State of Deepfakes"
- Graphika, "A Revealing Picture", December 8, 2023
- McAfee, "Artificial Imposters"
- ITV News on the CCDH Grok report, January 23, 2026
- Internet Watch Foundation, 2025 Annual Data & Insights Report
- Thorn, "1 in 8 Teens Know Someone Targeted by Deepfake Nudes", March 3, 2025
- Ofcom, "A deep dive into deepfakes that demean, defraud and disinform", July 23, 2024
- University College London, August 2, 2023
- AI Incident Database, Incident 634
- US FTC, "Complying With the Take It Down Act", May 8, 2026
- Clifford Chance, Data (Use and Access) Act, February 6, 2026
- Regulation (EU) 2024/1689 (AI Act)