[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-method-scrubs-banned-concepts-from-ai-video-without-blur":10,"sections":44},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":34,"tags":35,"sources":39,"feedback":43,"feedback_at":22,"cost_usd":43,"total_tokens":43},7012,"new-method-scrubs-banned-concepts-from-ai-video-without-blur","New Method Scrubs Banned Concepts From AI Video Without Blur","CleanVideo removes specific objects or themes from AI-generated video frame by frame without wrecking the rest of the clip.","A new technique called CleanVideo can strip banned or unwanted content out of AI-generated video without leaving the rest of the clip warped or jittery.\n\nThe method, described in a new arXiv paper, targets text-to-video diffusion models, the systems behind tools that turn a written prompt into a moving clip. Instead of applying one blanket fix, CleanVideo uses a tri-modal gating system that reads the video's spatial and temporal features, the diffusion model's timestep, and the text prompt itself to decide exactly where, when, and whether to intervene. When a clean substitute concept exists, it steers the banned content toward that substitute rather than just deleting it outright. Tested on three separate video diffusion models, the approach beat existing erasure baselines on both frame-by-frame and whole-clip evaluations, and it held up even when attackers tried to coax the erased concept back out.\n\nVideo is a harder erasure problem than static images because a concept can drift in and out of frame, change appearance across the clip, and shift with each denoising step, so a fix tuned for one frame can fail on the next. That gap is why concept removal in image generators has moved faster than in video models, and why platforms shipping video generation to the public need something more precise than a single global filter.\n\nEarlier attempts at scrubbing concepts from video often left behind a visible blur or stutter every time the model tried to censor itself, trading one flaw for another. CleanVideo's targeted approach suggests that tradeoff is not inevitable, though the real test will be whether it holds up against concepts nobody thought to name in the paper.","[\"ai\",\"text-to-video\",\"content-moderation\",\"diffusion-models\"]","2026-09-18T04:00:00.000Z","2026-09-19T06:41:26.884Z","2026-09-19T06:41:38.780Z","published",null,[24,30],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Attribute the findings to their actual source—this is an unreviewed arXiv preprint (arXiv:2609.20267) with no named authors or institution, so state that explicitly (e.g. 'in a paper posted to arXiv') instead of the vague 'a group of researchers,' and note it hasn't been peer reviewed.","resolved",{"id":31,"reviewer":26,"round":32,"reason":33,"status":29},"editor-r2",2,"Fix the garbled, nonsensical clause 'a chores of visible blur every time a model tries to censor itself' at the end of the context paragraph into a coherent sentence.","ai",[34,36,37,38],"text-to-video","content-moderation","diffusion-models",[40],{"name":41,"url":42},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.20267",0,{"sections":45},[46,50,55,60,65,70,75,80,84,89,94,99,104,109],{"name":47,"slug":34,"count":48,"latest_published_at":49},"AI",4114,"2026-09-19T18:33:46.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Security","security",673,"2026-09-19T17:30:00.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Policy","policy",345,"2026-09-19T19:22:20.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Hardware","hardware",156,"2026-09-19T11:00:00.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",129,"2026-09-19T19:45:00.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Consumer Tech","consumer-tech",99,"2026-09-09T17:27:33.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":18},"Dev Tools","dev-tools",78,{"name":85,"slug":86,"count":87,"latest_published_at":88},"Software","software",75,"2026-09-10T20:41:21.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"Startups","startups",55,"2026-09-09T23:14:29.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Gaming","gaming",43,"2026-09-10T12:18:06.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"General","general",42,"2026-09-18T22:35:10.000Z",{"name":105,"slug":106,"count":107,"latest_published_at":108},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":110,"slug":111,"count":112,"latest_published_at":113},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]