Deepfakes: Factsheet 2024-25

Blog post description.

7/24/20252 min read

a blurry image of a man's face in front of a wall with
a blurry image of a man's face in front of a wall with
Deepfakes 2024–2025: Key Facts & Figures

Prevalence & Growth

  • 10x increase in global deepfake incidents detected across all sectors in the past year.

  • 3000% rise in AI-powered fraud attacks from 2022 to 2023.

  • 700% surge in deepfake-enabled financial scams (2023).

  • 60% of consumers globally have encountered a deepfake video in the past year; only 15% say they have never seen one.

  • 95% of all deepfake videos made using open-source DeepFaceLab software.

Main Threats & Impact Areas

  • 88% of deepfake fraud targets the cryptocurrency sector.

  • 96% of deepfake videos online are pornographic/non-consensual.

  • 80% increase in deepfake-related cybercrime in South and East Asia in the last year.

  • 400+ companies per day targeted by deepfake audio/video CEO scams (USA).

  • $500,000: Average cost of a deepfake scam for affected businesses (2024).

Notable Incidents (2024–2025)

  • $25 million lost: Arup (Hong Kong) – deepfake Zoom scam with AI-cloned CFO and team.

  • $620,000 lost: Inner Mongolia businessman scammed by AI video call (China).

  • Pentagon fake explosion image: Caused real-world panic and stock market dip (USA, 2023).

  • US Election: AI-voice of Joe Biden used in robocalls urging voters not to vote.


AI Voice Cloning

  • 3 seconds of audio needed to clone a voice with ~85% accuracy.

  • 1 in 10 people report receiving a scam call using an AI-cloned voice of a loved one.

  • 77% of those targeted by AI voice scams lost money.


Detection Challenges

  • 51% accuracy: Average human ability to detect AI-generated images (no better than random guess).

  • 65% success rate: Best AI detection algorithms against advanced deepfakes.

  • 704% increase in deepfake attacks on ID verification systems (2023).


Nepal-Specific Data

  • 68% of Nepali youth aware of generative AI; 46% actively use AI tools (2024).

  • 635 cybercrime cases involving children (2023–24), up 260% from the previous year; most involve AI-manipulated images.

  • No current law specifically criminalizes deepfakes in Nepal (as of 2025).


Future & Security Risks

  • By 2027: Up to $40 billion in annual global losses from AI-driven fraud (U.S. estimate).

  • 30% of enterprises by 2026: Standard identity verification controls predicted to be insufficient due to deepfakes.

  • Real-time deepfake impersonation now possible in live video/voice calls.

  • Widespread “liar’s dividend”: Growing public skepticism toward all visual and audio evidence.


Source: Compiled from global cybersecurity reports, World Economic Forum, Reuters, Time, local Nepali media, UNICEF/ChildSafeNet studies, and expert analyses (2024–2025).