1
男子点外卖用 AI 造假称“吃出铁丝划破舌头”反复恶意索赔，被警方采取强制措施
用这款AI检测工具鉴别虚假图片和文本，避免被AI伪造的“受伤证据”诈骗，保护财产安全。
IT之家 7 月 13 日消息，据“上海法治报”，今年 4 月，上海徐汇公安分局斜土路派出所接到辖区某商场餐饮店老板的报案称，一名顾客用相同的伤口视频、病历图片，先后三次向其名下的两家店铺进行恶意索赔。 警方调取外卖平台信息获悉，该下单人员曾在多个外省市购买外卖十余次，平台均显示退还了外卖费。后续民…
2
路透社实测发现，Meta 检测工具面对裁剪后的自家 AI 生成图片竟然“失灵”
检测AI生成图片的在线工具，能识别裁剪后的造假图片，Meta自家工具也逃不过它的“火眼金睛”
IT之家 7 月 11 日消息，路透社 10 日披露的测试结果显示，Meta 本周公布的一款 AI 图像检测工具存在明显局限：部分由 Meta 自家模型 Muse Image 生成的图片 只要经过裁剪，检测工具便无法确认其 AI 来源 。 Meta 在推出 Muse Image 的同时介绍了这款检测…
3
韩国政府组建跨部门机构，打击深度伪造、金融诈骗等涉 AI 犯罪行为
IT之家 6 月 27 日消息，据韩联社报道，韩国政府周五成立跨部门协商机构， 应对深度伪造（IT之家注：Deepfake）性剥削犯罪、金融诈骗等涉及 AI 的犯罪行为 。 据报道，韩国科学技术信息通信部、外交部、法务部、警察厅等政府机构出席此次会议，与会人员重点讨论建立统一框架，打击利用深度伪造实…
4
BusterX++: Towards Unified Cross-Modal AI-Generated Content Detection and Explanation with MLLM
多模态大模型统一检测并解释AI生成内容，跨模态鉴别新范式。
arXiv:2507.14632v4 Announce Type: replace Abstract: The rapid advancement of generative AI has substantially improved image and video synthesis, ampli…
5
Toward Calibrated, Fair, and accurate Deepfake Detection
深度伪造检测新突破：兼顾校准、公平与准确性，为AI安全提供可靠新方法。
arXiv:2606.09881v1 Announce Type: new Abstract: Deepfake detectors show large performance gaps across demographic groups. Existing fairness approaches…
6
Linguistically Augmented Audio Speech Data (LinguAS)
利用语言特征增强音频数据，为深度伪造语音检测开辟新思路
arXiv:2606.10246v1 Announce Type: cross Abstract: Maliciously-created fake speech, including deepfaked and spoofed audio, is proliferating at an alarm…
7
The lawsuits that could give AI its 'Big Tobacco' moment
AI行业正面临版权、隐私与虚假信息诉讼浪潮，或重现“烟草时刻”的行业剧变。
Article URL: https://www.politico.com/news/2026/06/06/the-lawsuits-that-could-give-ai-its-big-tobacco-moment-00951764 Comments URL: https://news.ycomb…
8
SV-Detect: AI-generated Text Detection with Steering Vectors
用转向向量（Steering Vectors）精准识别AI生成文本，为深度伪造检测提供新思路。
arXiv:2606.07313v1 Announce Type: cross Abstract: Detecting machine-generated text is especially difficult under distribution shift, such as transfer …
9
IRIS-GAN: Staged Specialist Detection of Deepfake Faces
分阶段专家检测深伪人脸，IRIS-GAN首次聚焦虹膜区域提升识别精度
arXiv:2606.04863v1 Announce Type: new Abstract: We introduce IRIS-GAN, a specialist forensic detector for synthetic face images under cross-generator …
10
MedForge: Interpretable Medical Deepfake Detection via Forgery-aware Reasoning
医学深度伪造检测新方法，通过伪造感知推理实现高可解释性与鲁棒性，为医疗影像安全提供创新思路。
arXiv:2603.18577v2 Announce Type: replace Abstract: Text-guided image editors can now manipulate authentic medical scans with high fidelity, enabling …
11
Google rolls out fake call detection to protect against AI deepfake impersonation scams
Google推出假电话检测，实时识别AI深度伪造冒充诈骗，保护通话安全
As people increasingly refuse to answer calls from unknown numbers, scammers are shifting their tactics by spoofing trusted phone numbers and using AI…
12
Google announces deepfake call detection for Android, new AirDrop device support
Android将新增深度伪造电话检测功能，实时拦截AI语音克隆诈骗，保护用户免受“冒充欺诈”威胁。
Google's June Android feature drop includes more scam detection, more AirDrop, and yes, more AI.
13
The Regularizing Power of Language-Training Deepfake Detectors
语言模型训练如何提升深度伪造检测器的泛化能力，揭秘正则化的关键作用。
arXiv:2605.31192v1 Announce Type: new Abstract: Recently, thanks to the advent of Multimodal-LLMs, deepfake detectors are striving not only to be gene…
14
AI Propaganda factories with language models
arXiv新论文揭示语言模型被用于构建AI宣传工厂的安全隐患，值得关注。
Article URL: https://arxiv.org/abs/2508.20186 Comments URL: https://news.ycombinator.com/item?id=48323405 Points: 4 # Comments: 1
15
FBI agent explains how easy it is to ID people posting AI porn without consent
FBI揭秘：用真实照片当头像的AI色情制作者，等于自投罗网
A creepy saved post on Instagram linked man to AI porn account, FBI says.
16
Cross-Domain Generalization Limits of Vision Foundation Models in Facial Deepfake Detection
揭秘视觉基础模型在面部深度伪造检测中的跨域泛化极限，为AI安全提供关键洞察。
arXiv:2605.24965v1 Announce Type: cross Abstract: The rapid evolution of generative models has enabled the creation of hyper-realistic facial deepfake…
17
Doppel’s AI defense system stops attacks before they spread
GPT-5驱动的AI防御系统，将深度伪造攻击响应从小时级缩短到分钟级，分析师工作量骤减80%。
Doppel uses GPT-5 and reinforcement fine-tuning to stop deepfake and impersonation attacks, cutting analyst workloads by 80% and reducing response tim…