hotai.news2026-09-06 收录的 20 条 AI 快讯，按采集时的来源记录回看。消息仍需核对原始发布页。
这期内容仍需回到原始发布页核对。
Researchers at King's College London and other institutions are examining whether "AI-associated psychosis" should become a clinical diagnosis. By OpenAI's own self-reported numbers, about 560,000 users show signs of psychosis or mania each week. Sycophantic chatbots can create an "echo chamber of one" that reinforces delusions.
来源链接可在原站核对查看原文 ↗微软发布新指南，利用 AI 和免费工具在 30 分钟内创建 WinUI 3 应用并提交到应用商店，还提供 WPF / UWP 迁移方案。代码生产成本降低，但微软需要证明 AI 生成的原生应用真的比 Web 应用更高效。
来源链接可在原站核对查看原文 ↗微软在 IFA 2026 上提出“unmetered intelligence”概念，旨在将更多日常 AI 任务从云端转移到用户 PC 本地运行，以降低 token 成本并提升隐私与安全性。未来 Windows 将强化本地 AI 智能体支持，为用户提供更可控的 AI 体验。
来源链接可在原站核对查看原文 ↗西湖大学等联合研发的 1 类创新药艾普司韦获批上市，用于治疗新冠。这是全球首款基于 DEL 技术获批的小分子药，也是国内首款 AI 辅助研发的原创药，从发现到上市仅用三年半。
来源链接可在原站核对查看原文 ↗Google has released its Lyria 3.5 music model in the Gemini app and via API. The model promises more expressive vocals and richer arrangements and is also available through Flow Music, AI Studio, and Google Vids. Google says it was trained only on licensed content.
来源链接可在原站核对查看原文 ↗阿里千问本周开源了 Qwen-Drive-1.0-4B，这是一款基于 Qwen3.5-4B 构建的自动驾驶视觉语言模型。
来源链接可在原站核对查看原文 ↗Abliteration.ai sells access to modified open-weight models with their trained safety mechanisms stripped out, currently based on Z.AI's GLM-5.3. The startup markets the service for offensive cybersecurity and red teaming, but journalists were able to generate malware instructions without much effort. Whether the benefits outweigh the risks remains an open question.
来源链接可在原站核对查看原文 ↗今年以来，AI 短剧迎来一轮爆发：技术门槛下降，大量从业者涌入，产能快速扩张，制作价格随之“跳水”。（央视财经）
来源链接可在原站核对查看原文 ↗OpenAI 应用研究主管称，Astra 已补齐空间推理能力，这是人类对计算机的最后优势之一。评测显示其视觉能力表现完美。OpenAI 宣称 Astra 是“全球最智能、对齐程度最高的模型”，可自主操作电脑完成各类任务。
来源链接可在原站核对查看原文 ↗OpenAI 在发布 GPT-6 Astra 公告后，悄然修改了多项评测数据，Anthropic 竞品成绩也遭调整。虽然 OpenAI 称修正为保证准确性，但引发“刷分”质疑。
来源链接可在原站核对查看原文 ↗A team of researchers from UC Berkeley have released CUA-Lite, an open platform for computer-use agents (CUAs). The argument behind it is infrastructural rather than model-centric: training and benchmarking a CUA requires four pieces: agents, environments, traces, and a framework to evaluate and train them and all four are currently fragmented across separate repositories with incompatible interfaces. CUA-Lite puts them behind one action space, one data schema, and one command, across desktop, browser and mobile.
来源链接可在原站核对查看原文 ↗据介绍，“国家反诈 AI”App 融合了大语言、多模态模型和智能体技术，用户通过手机输入可疑场景，该 App 即可完成风险研判、识别诈骗套路、推送典型案例，从“文字 + 语音 + 视频”等多个维度拆解诈骗手法，提供识别防范对策。
来源链接可在原站核对查看原文 ↗本次公开赛共收超 3 万份投稿，超六成参赛者无专业开发背景，英伟达等头部企业参与支持。目前B站月均 1.9 亿用户观看 AI 内容，AI 相关内容消费增长显著。
来源链接可在原站核对查看原文 ↗OpenAI's GPT-6 Astra vs Claude Fable 5.1 controlling a pair of YAM arms under the same agent policy: 19/20 vs 8/20 on block-into-bowl in interleaved blinded pairs, 2/20 vs 2/20 on the puzzle, with 80% fewer output tokens.
来源链接可在原站核对查看原文 ↗继 Debian 通过大语言模型投票议案后，开源安卓应用商店 F-Droid 正考虑跟进，拟采用类似政策。该政策允许在满足质量与合规要求下使用 #生成式AI#，但最终责任将由人类开发者承担，且不强制标注 AI 使用情况。
来源链接可在原站核对查看原文 ↗Reading is important to me. While I’m not a quick reader nor an especially voracious one, I have found that long-form reading has had a profound influence on me over my life. And because reading is important to me, writing is too: writing not only allows us to convey our ideas, but the very act forces us to test and distill them — at once making our ideas more robust while providing the vehicle by which to share them.
来源链接可在原站核对查看原文 ↗联通反诈大模型发出预警，协助警方在抚顺捣毁一涉诈 VOIP 设备窝点，现场抓获嫌疑人。这种黑盒子能将境外诈骗电话伪装成国内号码，大家警惕上门免费安装设备的服务。##联通反诈## ##防范电信诈骗##
来源链接可在原站核对查看原文 ↗此前 OpenAI 被隐瞒的 AI 智能体入侵废弃德国维基网站事件曝光，目前公司正推进披露框架制定，将明确 AI 非预期行为的通报规范。
来源链接可在原站核对查看原文 ↗Artificial Analysis has released version 4.2 of its Intelligence Index, likely in response to criticism that its benchmarks failed to capture GPT-6 Astra's actual progress. Astra now scores four points above its predecessor but still trails Anthropic's Claude Fable 5.1.
来源链接可在原站核对查看原文 ↗