The EU AI Act rules that require deepfakes to be labelled apply from August 2026.
Why this market is moving now
- Verified deepfake attacks reached 821 in the first half of 2026 alone.
- The EU AI Act transparency rules, which require deepfakes to be labelled, apply from August 2026.
- Fraud is shifting to more advanced methods, with sophisticated identity fraud up 180% in 2025.
- Covered US platforms must remove reported intimate deepfakes within 48 hours under the TAKE IT DOWN Act.
Key figures
Global figures for this market, with the year each one describes. Use Cite to copy a reference to this page.
| Figure | Value | Year | Cite |
|---|---|---|---|
| In the first half of 2026, 821 verified deepfake attacks were documented worldwide, drawn from 1,760 news reports. | 821 attacks | 2026 | |
| About 1 in 6 documented deepfake attacks in the first half of 2026, or 137 incidents, involved nonconsensual sexual imagery of adults or children. | 137 incidents | 2026 | |
| The global identity fraud rate was 2.2% in 2025, down from 2.6% in 2024 but above 2.0% in 2023. | 2.2 % (global identity fraud rate) | 2025 | |
| Sophisticated identity fraud worldwide rose 180% in 2025 compared with 2024. | 180 % increase | 2025 | |
| In a 2026 global survey, 73% of respondents had been affected, or had someone in their network affected, by cyber-enabled fraud during 2025. | 73 % of respondents (global) | 2025 | |
| 55% of chief information security officers polled at a 2024 international cybersecurity meeting rated deepfakes a moderate to significant threat to their organization. | 55 % of CISOs polled | 2024 | |
| Trade in deepfake tools on dark web forums rose 223% between the first quarter of 2023 and the first quarter of 2024. | 223 % increase | 2024 | |
| In 2024 a Hong Kong employee of engineering firm Arup paid about US$25.6 million (HK$200 million) to fraudsters after a video call with deepfake re-creations of colleagues. | 25.6 USD million | 2024 | |
| The EU AI Act requires deepfakes to be clearly and visibly labelled, and its transparency rules apply from August 2026. | August 2026 rules apply from | 2026 | |
| The US TAKE IT DOWN Act, approved in May 2025, requires covered platforms to remove reported non-consensual intimate images, including deepfakes, within 48 hours of a valid request. | 48 hours | 2025 |
Calculated figures (shares, changes and prices per kilogram) are plain arithmetic on the reported values. Read how we work.
Companies to watch
Reality Defender
Deepfake detection platform for voice, video and images; KPMG LLP took a minority equity stake in the company. See
Pindrop
Sells real-time detection of synthetic voices and AI voice agents for contact centers and enterprises. See
GetReal Security
Sells real-time deepfake detection and continuous identity verification for calls, meetings and hiring. See
Resemble AI
Sells deepfake detection for audio, video and images through APIs and a Microsoft Teams integration. See
What the full report covers
- 01Executive summary
The answer in one page: size, direction and what moves it - 02Market definition and scope
What is in and out of the global market - 03Global market size, 2021 to 2025
Historical series and growth rates - 04Countries and regions
Demand, supply and trade by country and region - 05Outlook to 2031
Base, low and high scenarios with stated assumptions - 06Rules, policy and deadlines
Regulation and programmes that change demand - 07Drivers and barriers
What accelerates or slows adoption, by customer segment - 08Prices, costs and margins
Unit economics where public data allows - 09Competitive landscape
8 to 12 company profiles with recent moves - 10Opportunities and risks
For new entrants, suppliers and investors - 11Method and assumptions
How every figure was built
Included with the full report
- PDF report, 40 to 60 pages
- Excel data pack
- Ready-to-use charts
- Method note and assumptions
- Written to order in 5 working days
Frequently asked questions
How big is the global deepfake detection market?
There is no sourced global market size yet. As a measure of demand, 821 verified deepfake attacks were documented worldwide in the first half of 2026, and sophisticated identity fraud rose 180% in 2025.
How common is identity fraud worldwide?
The global identity fraud rate was 2.2% in 2025, down from 2.6% in 2024. In a 2026 global survey, 73% of respondents had been hit, or had someone in their network hit, by cyber-enabled fraud in 2025.
What is a well-known deepfake fraud case?
In 2024 a Hong Kong employee of engineering firm Arup paid about US$25.6 million to fraudsters after a video call in which colleagues were deepfake re-creations.
What rules apply to deepfakes around the world?
The EU AI Act requires deepfakes to be clearly labelled, with transparency rules applying from August 2026. In the US, the TAKE IT DOWN Act requires covered platforms to remove reported intimate deepfakes within 48 hours.
How worried are security chiefs about deepfakes?
55% of security chiefs polled at a 2024 international meeting rated deepfakes a moderate to significant threat to their organization. Trade in deepfake tools on dark web forums rose 223% between early 2023 and early 2024.
What kinds of deepfake attacks are most common?
About 1 in 6 verified deepfake attacks in the first half of 2026, or 137 incidents, involved nonconsensual sexual imagery of adults or children.
Free sample
Get a free sample of this report
We send the table of contents, two sample chapters and the method note, usually within one working day. It is free to ask.
- Full table of contents
- Two sample chapters
- Method note and data outline