Research map / Detection and segmentation / CNN features
GODDS: The Global Online Deepfake Detection System
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Describes GODDS, a publicly available deepfake detection system for journalists that ensembles detectors, auto-derives subject context from open sources, and uses human analysts in the loop to issue reports.
Problem
Deepfake audio, video, and image artifacts are proliferating, and journalists need accurate, explained verification of high-profile media where a single detector's error is costly.
Approach
Combines an ensemble of deepfake detectors with automatically gathered contextual information about the subject (news, Reddit, WikiData, reverse image search) and a human analyst review step before a report is emailed to the journalist.
Key results
- GODDS has been used by over 50 journalists from a dozen countries since its public release on July 8, 2024.
- Earlier versions of the underlying algorithms have been used since September 2023.
- Analysts view predictions from 20 audio deepfake detectors on a performance chart when reviewing an artifact.
Surprising findings
- Individual detectors can disagree sharply on the same audio (e.g., one predicting real while many predict fake), which is why the authors argue a single-detector report would be unreliable.
Limitations
- Context retrieval only works for high-profile subjects, events, and organizations where online context is available.
- Relies on human analysts in the loop, so throughput depends on staff.
- No quantitative accuracy evaluation reported; feedback described only qualitatively as positive.
Closest papers by meaning
- TriDF: Evaluating Perception, Detection, and Hallucination for Interpretable DeepFake Detection (CVPR 2026 · similarity 0.80)
- Through the Lens: Benchmarking Deepfake Detectors Against Moiré-Induced Distortions (NeurIPS 2025 · similarity 0.79)
- ODDN: Addressing Unpaired Data Challenges in Open-World Deepfake Detection on Online Social Networks (AAAI 2025 · similarity 0.79)
- FreqDebias: Towards Generalizable Deepfake Detection via Consistency-Driven Frequency Debiasing (CVPR 2025 · similarity 0.79)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · similarity 0.79)
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.79)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.