EviScene: Beyond "Is It AI?" for Agentic Image Factuality Assessment

Determining whether an image is AI-generated does not establish whether the situation it depicts is factually supported.

By Wang, Liu, Zhang +10Oct 10, 2026

Score█████░░░░░4.9

Key numbers

79.05% balanced accuracy and a

Caveats

Preprint; not yet peer reviewed.

VerdictCompetent work. Briefs at most.

Read the original PDF Code

Abstract

Determining whether an image is AI-generated does not establish whether the situation it depicts is factually supported. We formulate Image Factuality Assessment: given an image and a generic verification instruction, a system returns a supported or refuted verdict with an evidence-grounded explanation. Without a supplied textual claim, the system must progressively refine factual questions during investigation and align external evidence with the specific situation depicted. We introduce EviLens, an expert-reviewed benchmark of 10,174 supported and refuted images from controlled construction and web collection, covering twelve factual error types and evaluating both verdict accuracy and evidence sufficiency. We develop EviScene, a multimodal ReAct agent that interleaves visual inspection, external retrieval, and evidence analysis, using new findings to refine its questions and revisit relevant image details. We train a compact policy with supervised fine-tuning on reviewed teacher trajectories, followed by privileged on-policy self-distillation applied to verified decisions from the policy's own investigations. On the 1,527-image test set, the trained Qwen3.5-9B agent achieves 79.05% balanced accuracy and a Strict Evidence Sufficiency Rate (SESR) of 52.26%, improving over the same-backbone base agent by 6.71 and 9.37 percentage points, respectively. Its SESR is the highest among the evaluated agents.Code, models, and data will be released at https://github.com/wangzian828/EVISCENE and https://huggingface.co/ZiAnwang/EviScene.

ZiAn Wang, Ruiqi Liu, Zhengbo Zhang, Yi Han, Jialiang Shen, Ruoxin Chen, Manni Cui, Ximo Zhu, Jin Jiang, Mingzhe Liu, Linhao Wang, Wenhao Wang, Shu Wu

The editor's rubric

Heuristic review

Dimension	Level	Weight	What that level means
Leverage	███░░ 3	24%	A method or resource many groups across the field will adopt within a year.
Magnitude	██░░░ 2	18%	Solid incremental gain on a meaningful problem.
Evidence	███░░ 3	14%	Solid: multiple benchmarks or cohorts, ablations, fair baselines, released code or data.
Novelty	██░░░ 2	16%	A new combination of known ideas.
Trajectory	██░░░ 2	18%	Some room to improve with obvious engineering.
Stakes	██░░░ 2	10%	Benefits a professional community (practitioners, clinicians, engineers).

Editor’s rationale

Heuristic triage from title and abstract text only, not a reading of the paper. Cues found: method (we propose); verification (held-out test, code released); stakes (general AI). Red flags: derivative (comparative study).

How the score was computed

rank-2026-10-07

Score█████░░░░░4.9

Score = 10 × (65% × adjusted merit / 10 + 25% × attention + 10% × freshness)

Merit: 4.8 / 10; Weighted rubric, evidence-gated.
Adjusted merit: 4.3 / 10; Shrunk toward the desk prior by editor confidence (36%).
Attention: 72%; Citations, upvotes, points, mentions.
Freshness: 30%; Half-life decay since publication.

Hugging Face upvotes40 (reference 25, via hf-daily, Oct 14, 2026, 13:49 UTC)

The record

Reviewed by heuristic-v6 on Oct 14, 2026, 01:48 UTC. Paper type: method.
Categories: cs.CV
BRIEF, No.9 in the Artificial Intelligence edition of October 14, 2026.