VeridisQuo open-sources explainable deepfake detector

VeridisQuo stands out because it tackles both detection quality and interpretability instead of stopping at a fake-or-real score.

• –The hybrid spatial-plus-frequency design is well matched to compressed social video, where spectral artifacts can survive even when pixel clues get blurred away.
• –The explainability layer is the real product hook, since highlighting manipulated regions makes the detector easier to debug, demo, and trust.
• –Shipping code, inference endpoints, and a live demo gives developers a usable baseline for building moderation, forensics, or media verification workflows.
• –The main caveat is evaluation breadth: results on FaceForensics++ are useful, but cross-dataset testing on Celeb-DF or DFDC will matter more for real-world credibility.