AUDITS: A Comprehensive Benchmark for Image Manipulation Localization Across Multiple Analysis Axes
Researchers introduce AUDITS (Analysis Under Domain-shifts, qualIty, Type, and Size), a benchmark of over 530K images designed to evaluate image manipulation detection across multiple axes including domain shift, manipulation type, and size. The dataset draws from user and news photos and incorporates recent diffusion-based inpaintings. Experiments assess the robustness of existing manipulation detection methods under various domain shifts, aiming to advance development of more generalizable detection approaches.
Related guides (3)
Related events (8)
Automated Benchmark Auditing for AI Agents and Large Language Models (ABA)
The paper introduces Auto Benchmark Audit (ABA), an agentic framework that systematically audits AI benchmark tasks for issues such as ambiguous specifications, environment conflicts, and incorrect ground truths. Applied to 168 benchmarks across nine domains including NeurIPS publications, ABA identifies critical issues in over 25.7% of evaluated tasks. The authors demonstrate that filtering out flawed tasks materially shifts model rankings and improves average performance on SWE-bench Verified and Terminal-Bench 2 by 9.9% and 9.6% respectively, indicating that current benchmark scores are significantly distorted by task quality problems. The agentic tool and annotations are released publicly.
Multi-domain benchmark for detecting AI-generated text-rich images from GPT-Image-2
Researchers introduce a new benchmark of 8,602 images across six categories (commercial posters, infographics, academic posters, receipts, tables, UI screenshots) specifically for detecting AI-generated text-rich images produced by OpenAI's GPT-Image-2. Five zero-shot detectors are evaluated, revealing highly domain-dependent performance and severe sensitivity to JPEG compression even in the strongest conventional detector. A multimodal VLM is also explored as a detector, showing promise but limitations on structured formats. The work highlights a gap in existing benchmarks that focus on object-centric rather than text-layout-centric images.
OpAI-Bench: Benchmark for detecting AI text across progressive human-AI co-editing workflows
Researchers introduce OpAI-Bench, a benchmark for studying AI-text detection across progressive human-to-AI document revision workflows, covering document, sentence, token, and span granularities. Starting from human-written documents, the benchmark constructs nine sequentially revised versions per sample under five AI edit operations and varying AI coverage levels across four domains. Key findings include that mixed-authorship intermediate versions are often harder to detect than fully human or heavily AI-edited endpoints, revealing non-monotonic detection patterns absent from existing benchmarks. The work addresses a gap in AI-text detection research as real-world documents increasingly result from iterative human-AI co-editing rather than pure generation.
Social Gaze Consistency as a Semantic Cue for AI-Generated Image Detection
This paper introduces Social Gaze Consistency (SGC), a high-level semantic detection axis based on the mutual coherence of gaze direction, head-eye alignment, and pupil placement between interacting individuals in images. The authors construct a controlled diagnostic dataset with region-specific gaze perturbations and a Block-Compositional Caption Supervision scheme to train detectors without generator-fingerprint memorization shortcuts. Cross-architecture validation shows +3.7 pp improvement on the COCOAI Interaction subset when applied to FakeVLM, with gains transferring from a single inpainter (FLUX.1-Fill) to multi-generator suites. The work argues that diffusion models share a spectral weakness in periocular structure, making gaze coherence a robust, backbone-agnostic detection signal orthogonal to existing low-level artifact methods.
Bayesian audit framework for public AI evaluation archives challenges frontier model claims
A new arXiv preprint proposes a Bayesian inference and decision-audit framework for interpreting public AI evaluation archives (LiveBench, Open LLM Leaderboard v2, LMArena, GAIA, tau-bench) as longitudinal time series rather than terminal leaderboards. The paper demonstrates that a single terminal snapshot is compatible with multiple distinct performance histories, yielding ambiguous timing estimates for reaching capability ceilings. A candidate selection-aware frontier model is shown to fail synthetic recovery, objective-archive prediction, preference transfer, and uncertainty calibration, with fixed audit gates rejecting its stronger claims. The work proposes an archive-and-adjudication protocol to reconstruct evaluation histories and falsify unsupported frontier capability claims.
KLIP: Localized OOD Detection in Inverse Problems via KL-Divergence with Diffusion Priors
KLIP proposes an out-of-distribution detection metric for computational imaging that computes KL-divergence between a diffusion model prior and the posterior distribution. Unlike prior approaches, it requires no calibration data or knowledge of the shifted distribution, and can both flag whole images and localize OOD patches within images. The method is validated on medical imaging tasks such as detecting liver tumors in CT scans and generalizes across diffusion model architectures, datasets, and inverse problem types.
WSADBench: A Unified Benchmark for Weakly Supervised Anomaly Detection
WSADBench is the first benchmark to unify evaluation across the three primary weakly supervised anomaly detection (WSAD) paradigms—incomplete, inexact, and inaccurate supervision—testing 36 algorithms across 4 modalities with over 700K experiments. Key findings challenge the isolation of current WSAD research directions, showing strong correlations between supervision scenarios and that specialized WSAD methods are quickly outperformed by tabular foundation models as label availability increases. The benchmark also reveals inconsistent utility of unlabeled data and asymmetric model sensitivity to label noise types. Code and datasets are released open-source.
StylisticBias benchmark reveals a small set of visual cues drives most social bias in MLLMs
Researchers introduce StylisticBias, a controlled benchmark of ~25K photorealistic face images with single-attribute variations designed to isolate how specific visual cues shift social judgments in multimodal LLMs. Evaluating six MLLMs across 25 binary social judgment scenarios, they find that age and body type dominate identity-level effects, while fashion style drives the largest attribute-level shifts, with ~15 attributes accounting for ~80% of total bias variation. The benchmark is released publicly on GitHub and Hugging Face, enabling fine-grained bias auditing of multimodal models.