Skip to content

๐Ÿ”Ž AIGC Detection

๐Ÿง  NeurIPS2026 ยท 14 paper notes

๐Ÿ“Œ Same area in other venues: ๐ŸŽž๏ธ ECCV2026 (8) ยท ๐Ÿ“ท CVPR2026 (10) ยท ๐Ÿ”ฌ ICLR2026 (30) ยท ๐Ÿ’ฌ ACL2026 (17) ยท ๐Ÿงช ICML2026 (11) ยท ๐Ÿค– AAAI2026 (2)

๐Ÿ”ฅ Top topics: LLM ร—3

BIAS-ID: A Framework for Analyzing Transformation Biases in AI-Generated Image Detectors

Introduces the BIAS-ID analysis framework, which systematically quantifies directional transformation biases in AI-generated image detectors by directly analyzing the signed "score shift" of predicted probabilities and the aggregated transform sensitivity (ATS) metric.

BIAS-ID: A Framework for Analyzing Transformation Biases in AI-Generated Image Detection

Addressing the frequent real-world collapse of AI-generated image detectors, this paper introduces BIAS-ID, an evaluation and quantification framework that decouples systematic transformation biases from general robustness decay via score shift and Aggregated Transform Sensitivity (ATS).

BusterX: MLLM-Powered AI-Generated Video Forgery Detection and Explanation

This paper introduces GenBuster-200K, a fair dataset of over 200K high-resolution videos, GenBuster-Bench with progressive dual-axis evaluation, and BusterX, an MLLM baseline trained via entropy-collapse-resistant reinforcement learning (DAPO) that turns long-chain spatiotemporal physical reasoning into the detector itself.

Can Pixels Alone Reveal Image Origin? Minimax Limits and Learnable Interfaces for Passive Provenance

The paper exactly characterizes the optimal robust separation available to pixel-only passive provenance verification through total variation distance, separates this statistical ceiling from interface information exposure in a deployed verifier, and validates the identity in a finite domain without establishing a robustness certificate for realistic images.

FARE: Forensic Acceptance Region Estimation for Catching Bait-and-Switch Image Generators

To tackle deployment-time bait-and-switch risks where black-box image generation APIs covertly replace certified models with cheaper substitutes, FARE constructs a forensic acceptance region from certified outputs alone via constrained residual extraction and boundary-tightening hard positives, achieving near-perfect swap detection under strict 1% false positive rates.

Hitting a Moving Target: Test-Time Adaptation for AI Text Detection under Continual Distribution Shift

Addressing the catastrophic failure of supervised AI text detectors under adversarial humanization, new model emergence, and human writing drift, this paper proposes leveraging inference-time homogeneity among unlabeled samples via positive-unlabeled test-time adaptation (TTA), substantially restoring detection and prevalence estimation performance.

LLM-Auction: Generative Auction towards LLM-Native Advertising

LLM-Auction introduces the first learning-based generative auction mechanism that unifies auction allocation and text generation into a single preference alignment framework, internalizing response-level externalities in a single forward pass while achieving state-of-the-art allocation efficiency and provable incentive guarantees.

Log-Likelihood, Simpsonโ€™s Paradox, and the Detection of Machine-Generated Text

Uncovers that likelihood-based machine-generated text detectors suffer from Simpson's paradox due to naive token-score aggregation across heterogeneous hidden-space regions, and introduces a lightweight, Bayesian-grounded local calibration framework that boosts Fast-DetectGPT AUROC on GPT-5.4 from 0.63 to 0.85+.

Majority Bit-Aware Watermarking for Large Language Models

To resolve the fundamental trade-off between watermarked text quality and decoding accuracy caused by small green list ratios in perturbation-based multi-bit LLM watermarking, this paper proposes majority bit-aware encoding (MajorMark and MajorMark+), which decouples watermark detectability from green list restriction, guarantees a large green list ratio \(\gamma \ge 0.5\), and significantly improves bit accuracy and text utility.

Post-hoc Selective Classification for Reliable Synthetic Image Detection

To fix the pathological behavior where conventional logit-based confidence scores assign higher confidence to misclassified synthetic images under covariate shifts, ReSIDe extracts layerwise logits via unsupervised linear probing on feature centroids and aggregates multi-layer confidence scores through pairwise preference optimization that provably minimizes an upper bound of the area under the risk-coverage curve (AURC).

Rethinking the Readout: Unlocking Video Backbones for AI-Generated Video Detection

Addressing the paradox where video pretrained backbones underperform image-based detectors on AI-generated video (AIGV) detection, this paper identifies standard spatiotemporal readouts (GAP/CLS) as the primary bottleneck due to inter-patch relation collapse and per-channel velocity cancellation, and proposes Velocity-Gated Patch Velocity Profiling (V-PVP)โ€”a lightweight ~0.5M-parameter dual-stream readout that reaches 95.28% AUC with a fully frozen backbone.

RiSE: Residual Subspace Expert for Generalizable Text-Centric Image Forgery Localization

RiSE prevents feature collapse by freezing principal SVD components of vision foundation models and adapting low-rank residual subspaces, constructs offline forgery-aware residual experts, and selects the optimal prediction via single-pass analytic Latent Perturbation Confidence (LPC), substantially advancing cross-domain text-centric forgery localization.

Steer-to-Detect: Probing Hidden Representations for Detection of LLM-Generated Texts

Steer-to-Detect (S2D) introduces a lightweight, learnable steering vector injected into an observer LLM's intermediate layers to reshape latent representations and maximize class separability, delivering highly accurate passive AI-text detection with finite-sample false-positive control.

VI-Bench: Benchmarking Prompt Inversion from AIGC Videos

Addressing the gap where video understanding benchmarks only evaluate descriptive recognition rather than generative control, this paper introduces VI-Bench, the first benchmark for prompt inversion from AIGC videos across five control dimensions, uncovering a severe divergence between textual prompt plausibility and generator replay fidelity across 18 frontier VLMs.