Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper redefines the detection of AI-generated video by leveraging the motion field encoded in the compressed bitstream, allowing for real-time analysis without the need for pixel decoding. By establishing a monotone aggregate scoring system, the method achieves significant computational efficiency, demonstrating a full-length AUC of 0.64 on GenVidBench with five orders of magnitude less compute than traditional pixel-based CNNs. Notably, the approach allows for deferral of detection on 15% of clips, improving accuracy from 0.75 to 0.78 while maintaining a target false-positive rate, showcasing the effectiveness of the reframed detection strategy.
Achieving a 15% deferral in AI-generated video detection boosts accuracy significantly while slashing compute costs by $7\times$ compared to traditional methods.
Detectors for AI-generated video are evaluated offline. A clip is decoded to pixels and scored once, increasingly by a large vision-language model. Detection, however, is deployed online. We recast the task as streaming perception and score the motion field the codec already wrote into the bitstream. Reading that field is a parse, not a pixel-domain forward pass. Because the running aggregate is monotone, one end-calibrated threshold is anytime-valid at the data-dependent decision time. Recalibrating at each prefix is not. Escalation is priced in closed form. A compute budget maps to a deferral window, on a frontier monotone exactly where the deferral condition holds. On matched GenVidBench the codec stage reaches full-length AUC 0.64 at five orders of magnitude less compute than a pixel CNN, on CPU. Its gate holds the stopping-time false-positive rate at target while the real data match its calibration, and drifts above it under distribution shift. Deferring 15% of clips lifts accuracy from 0.75 to 0.78 at $7\times$ less compute (paired: McNemar $p<10^{-6}$). The stage-1 ordering replicates on AIGVDBench. We introduce no new detector. The contribution is the reframing, two guarantees, and the measured frontiers. Code, configurations, and evaluation splits: https://github.com/KurbanIntelligenceLab/streamdet.