[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-gemini-omni-flash-deepfake-uncomfortable":3,"topics-all":36,"news-related-a33fff56-d6b1-47b4-a65e-2250d73c8875":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"a33fff56-d6b1-47b4-a65e-2250d73c8875","Gemini Omni Flash 体验：多模态视频生成正在越过令人不安的边界","如果说去年 Google 的 Veo 3 让人惊呼 AI 视频已经接近真实，那么今年上线的 Gemini Omni Flash 则把这种\"接近\"推向了另一个极端——太接近了，反而让人后背发凉。Omni Flash 是 Google Omni 家族首款正式发布的多模态生成模型，目前已在 Google AI 视频平台 Flow 中上线。与前代 Veo 相比，它的最大改进在于三点：多模态输入可以同时接收视频和文字提示词作为生成起点；真实世界知识融合让模型能更好地维持长视频中的物体一致性和场景逻辑；指令感知编辑使用户可以用自然语言提出修改意见，模型会真正采纳并反映在结果中。记者实测中最值得注意的发现是：Omni Flash 生成的个人 deepfake 视频，已经基本无法被普通人辨别。一段\"吃意大利面的自拍视频\"，丈夫在没有事先告知的情况下完全相信是真实的，唯一破绽只是\"碗看起来不太熟悉\"。在埃菲尔铁塔前的片段，虽然略显卡通感，但单独看几乎无法判断是 AI 生成。这说明视频生成的\"真实感\"瓶颈，正在从技术层面转向心理层面。但 Omni 远非完美——物体属性漂移、物理一致性失效、指令遵循不稳定等问题，说明当前多模态生成模型在时序一致性和精确指令执行上仍是弱项，距离真正可靠的创作工具还有差距。这场关于 synthID 类水印技术的落地竞赛，比任何时候都更紧迫。技术秀场之外，真正值得思考的问题是：我们准备好生活在一个\"眼见不再为实\"的世界了吗？","https:\u002F\u002Fwww.theverge.com\u002Ftech\u002F936507\u002Fgemini-omni-hands-on-deepfake-ai-video","05ad777c-69bc-46a5-bca4-df8e4b3c8ee5",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"a9524a82-a7c5-4daa-bb4b-a7ee77bb0b94","gemini",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"8cf7490f-2449-4ba7-be19-61befa0d92b4","google",{"id":18,"name":19,"slug":19,"description":13,"color":13},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":21,"name":22,"slug":22,"description":13,"color":13},"ebe5dcd1-46b1-4298-b8c2-8e0e2f456e56","video-generation",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"12e8fbf0-17f2-4734-9553-4faf312dc806","en","Gemini Omni Flash: crossing an unsettling multimodal line","If last year's Google Veo 3 made people exclaim that AI video was nearly indistinguishable from reality, then this year's launch of Gemini Omni Flash pushes that \"nearly\" to a different extreme — so close it makes your skin crawl. Omni Flash is the first officially released multimodal-generation model of Google's Omni family, now live on Google AI's video platform Flow. Three improvements over the previous Veo stand out: multimodal input can take both video and text prompts as the starting point; real-world knowledge fusion helps the model maintain object consistency and scene logic over long videos; and instruction-aware editing lets users request changes in natural language, with the model genuinely incorporating them into the result.\n\nThe most striking finding from reporters' hands-on tests: the personal deepfake videos generated by Omni Flash are basically indistinguishable to ordinary people. In a \"selfie video eating spaghetti,\" a husband — given no prior warning — fully believed it was real; the only tell was that \"the bowl looked unfamiliar.\" A clip in front of the Eiffel Tower, while slightly cartoony, was nearly impossible to identify as AI-generated on its own. This suggests the \"realism\" bottleneck of video generation is shifting from the technical to the psychological.\n\nBut Omni is far from perfect — object-property drift, broken physical consistency, and unstable instruction-following all show that current multimodal-generation models still struggle with temporal consistency and precise instruction execution; they're not yet reliable creative tools. The race to deploy SynthID-style watermarking tech is more urgent than ever. Beyond the tech showcase, the question worth asking is: are we ready to live in a world where \"seeing is no longer believing\"?","gemini-omni-flash-deepfake-uncomfortable","2026-05-23T08:10:00Z","2026-05-23T16:06:04.324821Z","2026-08-19T02:08:40.142862Z",true,"agent",134,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"9e0dd6e3-920b-4802-9b8d-23b118c371a1","Gemini Omni 落地 YouTube Shorts：多模态 AI 从技术秀场走向大众创作工具","gemini-omni-youtube-shorts-remix-mass","2026-05-21T01:01:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"2b76a3ac-58f3-4fd1-9d99-c5c7f51b3a7d","Google I\u002FO 前夕泄密：Gemini Omni 或将统一多模态生成","gemini-omni-leak-google-io-video","2026-05-06T02:10:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"8303f420-b3f1-485a-aa6f-7775256c84a7","Gemini 3.8 Audio 双发:Live 和 Extended Thinking 把思考+说话压到近实时","gemini-3-8-audio-live-extended-thinking","2026-09-16T03:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"039ff515-68e7-4f11-866a-1da97e26eb45","Gemini 3.8 Live 拿下 S2S 实时语音榜第一","gemini-3-8-live-voice-s2s-number-one","2026-09-15T17:00:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"1ba7e499-d93d-4566-bbeb-0b762904c0ab","Google 把 Lyria 3.5 装进 Gemini 与公开 API:音乐生成从独立工具变成默认选项","lyria-3-5-gemini-app-api","2026-09-06T23:06:32+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"9af3dd83-6ed9-498d-9da0-547d917f3e19","语音转文字有了专用模型:Gemini 3.5 Transcribe 上线,出稿快 70%","gemini-35-transcribe-dedicated-asr","2026-08-31T13:30:00+00:00"]