Detect Early, Escalate Rarely: Anytime Detection of AI-Generated Video from the Compressed Bitstream

2026-07-21

AI Summary

Existing methods for detecting AI-generated videos rely on computationally expensive pixel-level models, which are ill-suited for online streaming applications. This work proposes the first bitstream-level detection framework that operates without full video decoding, directly extracting motion fields from compressed bitstreams and enabling valid early decisions at arbitrary time points through monotonic cumulative statistics. The approach provides two theoretical guarantees: detection thresholds valid at any time and closed-form control over false positive rates, while establishing a computation–latency Pareto frontier. On GenVidBench, it achieves an AUC of 0.64 with only 1/50,000th the CPU computation of a pixel-based CNN; further, by delaying processing on just 15% of samples, it significantly improves accuracy from 0.75 to 0.78 (p < 10⁻⁶) under 7× lower computational overhead.

Abstract

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.

Problem

• AI-generated video detection
• streaming perception
• compressed bitstream
• anytime detection
• online deployment

Innovation

• anytime detection
• compressed bitstream
• streaming perception
• motion field parsing
• compute-efficient AI detection