Deepfake-Eval-2024 benchmark
paperAuthors
Credibility Rating
Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.
Rating inherited from publication venue: arXiv
Relevant to AI safety practitioners concerned with misuse of generative AI for disinformation and fraud; highlights how rapidly deepfake technology outpaces detection, with implications for trust, verification, and governance of synthetic media.
Paper Details
Metadata
Abstract
In the age of increasingly realistic generative AI, robust deepfake detection is essential for mitigating fraud and disinformation. While many deepfake detectors report high accuracy on academic datasets, we show that these academic benchmarks are out of date and not representative of real-world deepfakes. We introduce Deepfake-Eval-2024, a new deepfake detection benchmark consisting of in-the-wild deepfakes collected from social media and deepfake detection platform users in 2024. Deepfake-Eval-2024 consists of 45 hours of videos, 56.5 hours of audio, and 1,975 images, encompassing the latest manipulation technologies. The benchmark contains diverse media content from 88 different websites in 52 different languages. We find that the performance of open-source state-of-the-art deepfake detection models drops precipitously when evaluated on Deepfake-Eval-2024, with AUC decreasing by 50% for video, 48% for audio, and 45% for image models compared to previous benchmarks. We also evaluate commercial deepfake detection models and models finetuned on Deepfake-Eval-2024, and find that they have superior performance to off-the-shelf open-source models, but do not yet reach the accuracy of deepfake forensic analysts. The dataset is available at https://github.com/nuriachandra/Deepfake-Eval-2024.
Summary
Deepfake-Eval-2024 introduces a large-scale benchmark of in-the-wild deepfakes collected from social media and detection platforms in 2024, revealing that state-of-the-art detectors suffer dramatic performance drops (45-50% AUC decrease) compared to academic benchmarks. The dataset spans 88 websites, 52 languages, and includes video, audio, and images using the latest manipulation technologies. Results show commercial and fine-tuned models improve over open-source baselines but still lag behind human forensic analysts.
Key Points
- •State-of-the-art open-source deepfake detectors drop ~50% AUC on real-world data vs. academic benchmarks, exposing a critical generalization gap.
- •Dataset includes 45 hours of video, 56.5 hours of audio, and 1,975 images collected from 88 websites in 52 languages in 2024.
- •Commercial and fine-tuned models outperform off-the-shelf open-source solutions but still cannot match the accuracy of human forensic analysts.
- •Academic benchmarks are shown to be significantly out of date and non-representative of modern deepfake generation technologies.
- •The benchmark is publicly available, enabling reproducible evaluation of deepfake detection systems against current real-world threats.
Cited by 2 pages
| Page | Type | Quality |
|---|---|---|
| Epistemic Collapse | Risk | 49.0 |
| AI-Powered Fraud | Risk | 69.0 |
Cached Content Preview
Deepfake-Eval-2024: A Multi-Modal In-the-Wild Benchmark of Deepfakes Circulated in 2024
Deepfake-Eval-2024: A Multi-Modal In-the-Wild Benchmark of Deepfakes Circulated in 2024
Nuria Alina Chandra
TrueMedia.org
Ryan Murtfeldt
TrueMedia.org
University of Washington, Seattle
Lin Qiu
TrueMedia.org
University of Washington, Seattle
Arnab Karmakar
TrueMedia.org
University of Washington, Seattle
Hannah Lee
TrueMedia.org
Emmanuel Tanumihardja
TrueMedia.org
University of Washington, Seattle
Kevin Farhat
TrueMedia.org
University of Washington, Seattle
Ben Caffee
TrueMedia.org
University of Washington, Seattle
Sejin Paik
TrueMedia.org
Georgetown University, Washington D.C.
Changyeon Lee
Miraflow AI
Yonsei University, Seoul
Jongwook Choi
TrueMedia.org
Chung-Ang University, Seoul
Aerin Kim
TrueMedia.org
Miraflow AI
Oren Etzioni
TrueMedia.org
University of Washington, Seattle
Abstract
In the age of increasingly realistic generative AI, robust deepfake detection is essential for mitigating fraud and disinformation. While many deepfake detectors report high accuracy on academic datasets, we show that these academic benchmarks are out of date and not representative of real-world deepfakes. We introduce Deepfake-Eval-2024, a new deepfake detection benchmark consisting of in-the-wild deepfakes collected from social media and deepfake detection platform users in 2024. Deepfake-Eval-2024 consists of 45 hours of videos, 56.5 hours of audio, and 1,975 images, encompassing the latest manipulation technologies. The benchmark contains diverse media content from 88 different websites in 52 different languages. We find that the performance of open-source state-of-the-art deepfake detection models drops precipitously when evaluated on Deepfake-Eval-2024, with AUC decreasing by 50% for video, 48% for audio, and 45% for image models compared to previous benchmarks. We also evaluate commercial deepfake detection models and models finetuned on Deepfake-Eval-2024, and find that they have superior performance to off-the-shelf open-source models, but do not yet reach the accuracy of deepfake forensic analysts. The dataset is available at https://github.com/nuriachandra/Deepfake-Eval-2024 .
Figure 1: Examples of Deepfake-Eval-2024 video and audio (rows 1–2), and images (rows 3–4), demonstrating a diversity of content styles and generation techniques, including lipsync, faceswap, and diffusion. Images have been resized for presentation.
1 Introduction
Advances in generative AI models have precipitated a surge of highly realistic deepfakes, which have been used to fabricate messages from politicians [ 1 ] , create non-consensual pornographic content [ 2 ] , spread misinformation [ 3 ] , and damage reputations [ 4 ] , harming lives, businesses, and nations [ 5 ] . Between 2023 and 2024, there was
... (truncated, 67 KB total)f39c2cc4c0f303cc | Stable ID: sid_bmdkuj7uLa