VideoShield: Adversarially Optimized Latent Space Injection for Architecture-Agnostic Video Diffusion Model Watermarking with Complete Dilution Curve Detection and Unified Proactive/Retroactive Invariant

We present VideoShield, a system for detecting unauthorized use of specific video works in video diffusion model training corpora, without requiring model weight access. Four contributions are made: (C1) Latent space injection — per-video cryptographic directional vectors injected into VAE latent tensors at LAT_STRENGTH=0.15, validated at 1% corpus dilution across 3 independent sessions (t = +997, p = 2.67e-46); (C2) Adversarially optimized DELTA_opt — gradient ascent produces detection at all dilution levels 1%-20%, eliminating two previously blind spots, with gains of 254x at 20% and 653x at 2%; (C3) Architecture-agnostic cross-VAE injection — restricting injection to ultra-low-frequency PSD bands 0-2 enables detection across 3 independent video VAE architectures: Wan 2.1 (Alibaba), CogVideoX (Tsinghua), HunyuanVideo (Tencent); (C4) Unified proactive/retroactive invariant — PSD bands 0-2 (DELTA_BB2) are simultaneously optimal for proactive watermarking (B6) and retroactive Membership Inference Attack detection (B7, t = 11,431, p = 8.28e-126), revealing that video DiTs preferentially memorize quasi-DC latent spatial components regardless of injection mechanism. A non-monotonic cross-contamination amplification effect is documented. The system generates ISO/IEC 27037-compliant forensic reports admissible under EU AI Act Article 53.v2 (March 30, 2026): Added citation and discussion of concurrent work by Hu et al. (arXiv:2501.14195, ICLR 2025) in Related Work (Section 2.4) and Discussion (Section 5.2). No changes to scientific content, experiments, or claims.