Summary
The second Deepfake Detection, Localization and Interpretability Challenge uses a multimodal deepfake description dataset enriched with Qwen3-VL. It evaluates more than image-level classification by targeting localisation and interpretable traces of manipulation.
Why it matters
Detection systems that only classify content as real or fake provide limited help for investigators and platforms. Localisation and interpretability are more useful for verification, remediation and accountability.
Related coverage
- TRUST-AI convenes on trustworthy AI
- AIM4DK frames disappearing knowledge as an AI reliability problem
- VALE advances value engineering as an AI research discipline