AI 能看出图片是假的，但看不出图里的事是假的
现在判断图片是不是 AI 生成，已经不够了：一张图可以是真人拍的，但图里的事是编的。这篇把问题升级成「图片事实核查」:不给你任何文字声明，只给一张图，让 AI 自己决定该查什么、怎么查，最后给出「有证据支持」或「被证伪」的结论，还要附上证据。研究者造了一个 1 万多张图的评测集，涵盖 12 类事实错误，并训练了一个 90 亿参数的小模型，在测试集上准确率 79%，证据充分率 52%，是目前同类里最高的。它不是你明天能拿来核查新闻图片的工具，但它把「AI 造假」从像素层面推进到了事实层面——以后判断一张图能不能信，标准会从「是不是 AI 画的」变成「图里的事是不是真的」。
📄 原文摘要(英文)
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.