[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-target-llm-safetys-worst-case-not-its-average":10,"sections":34},{"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":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},7741,"researchers-target-llm-safetys-worst-case-not-its-average","Researchers Target LLM Safety's Worst Case, Not Its Average","A new fine-tuning method caps rare severe safety regressions instead of just averaging risk, and it beats existing baselines in tests.","A new academic method wants to stop fine-tuning from quietly wrecking a model's safety behavior on its worst prompts, not just its average ones.\n\nResearchers describe a technique called chance-constrained fine-tuning, detailed in a paper posted to arXiv this week. Instead of tracking average safety loss across a whole evaluation set, the method limits the fraction of safety-critical prompts allowed to degrade past a set threshold compared with a reference model. Because that kind of pass\u002Ffail counting is not something gradient descent can optimize directly, the authors build a smooth stand-in for it and use that to steer fine-tuning updates so they stay inside a \"safe\" region of the model's parameters. They tested it on harmful fine-tuning scenarios across three tasks and three different models, and report it consistently beat existing safety-preserving baselines from the literature.\n\nThe framing is the real contribution here. Most current safety-preserving fine-tuning tools optimize an average, which is exactly the kind of metric a handful of catastrophic failures can hide behind if most other cases still look fine. Treating safety as a reliability constraint, where you cap how often things go badly wrong rather than how badly things go wrong on average, is a more honest description of what people actually want from a safety-tuned model.\n\nIt is one arXiv preprint, not a shipped safety pipeline, and \"chance-constrained\" still means a controlled failure rate, not zero failures. Whether any lab bothers to bolt this onto a real fine-tuning stack, rather than just writing another blog post about alignment, is the actual test.","[\"ai\",\"ai-safety\",\"fine-tuning\",\"research\"]","2026-09-25T04:00:00.000Z","2026-09-25T19:15:47.423Z","2026-09-25T19:15:52.453Z","published",null,[],"ai",[24,26,27,28],"ai-safety","fine-tuning","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.29960",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,90,95,100],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",4487,"2026-09-25T15:40:03.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",734,"2026-09-25T15:52:13.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",389,"2026-09-25T15:27:35.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",247,"2026-09-25T15:26:22.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",185,"2026-09-25T15:00:22.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Science","science",139,"2026-09-25T11:55:23.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Consumer Tech","consumer-tech",132,"2026-09-25T15:30:00.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",88,"2026-09-24T23:06:55.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",81,"2026-09-25T09:59:40.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Startups","startups",74,"2026-09-25T14:05:04.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"Gaming","gaming",47,"2026-09-25T13:33:16.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"General","general",46,"2026-09-25T02:12:57.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Reviews","reviews",30,"2026-09-24T20:07:31.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]