[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-study-maps-how-training-one-value-shifts-ai-behavior-elsewhere":10,"sections":40},{"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":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},11065,"study-maps-how-training-one-value-shifts-ai-behavior-elsewhere","Study Maps How Training One Value Shifts AI Behavior Elsewhere","A study of 66 alignment values finds that model activations predict behavioral spillover far better than text descriptions of values.","Researchers have a new way to predict how training an AI on one value warps its behavior everywhere else.\n\nAI labs fine-tune models to follow a list of values and traits, known as an alignment target, then test the model against that same list. The problem: training on those narrow, listed behaviors also reshapes how the model acts in situations nobody tested for, and those side effects are hard to anticipate. A team of researchers analyzed this spillover across 66 values pulled from alignment targets used in current models, building what they call a generalization matrix. They then tried to predict that matrix two ways: by reading the model's internal activations when it applies a value, and by analyzing plain-text descriptions of the values. The activation-based method hit a 0.45 correlation with actual generalization patterns; the text-based method managed only 0.05.\n\nThat gap matters because alignment evaluations mostly check whether a model nailed the behaviors it was explicitly trained on, not what else shifted. A method that forecasts the untested fallout, before a model ships, is more useful than finding out after release that a safety tweak nudged unrelated behavior. The researchers also used these activation-based representations to measure how similar the values within one alignment target are to each other, and found that similarity score tracked with how robust the resulting model was.\n\nCall it a map of AI's blind spots - useful, but a 0.45 correlation means plenty of territory is still unmapped.","[\"ai-alignment\",\"llm-safety\",\"machine-learning\"]","2026-10-09T04:00:00.000Z","2026-10-10T04:30:19.971Z","2026-10-10T04:30:25.097Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Fix the numeric inconsistency where the body claims evaluations miss 'hundreds of behaviors' when the study only covers 66 values and the headline\u002Fdek themselves say 'dozens' (65) — use a count consistent with the source.","resolved","ai",[32,33,34],"ai-alignment","llm-safety","machine-learning",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.12410",0,{"sections":41},[42,45,49,54,59,63,67,72,77,82,86,91,96,101],{"name":43,"slug":30,"count":44,"latest_published_at":18},"AI",6783,{"name":46,"slug":47,"count":48,"latest_published_at":18},"Security","security",932,{"name":50,"slug":51,"count":52,"latest_published_at":53},"Policy","policy",486,"2026-10-08T22:40:11.000Z",{"name":55,"slug":56,"count":57,"latest_published_at":58},"Deals","deals",474,"2026-10-08T22:00:00.000Z",{"name":60,"slug":61,"count":62,"latest_published_at":18},"Hardware","hardware",232,{"name":64,"slug":65,"count":66,"latest_published_at":18},"Science","science",193,{"name":68,"slug":69,"count":70,"latest_published_at":71},"Consumer Tech","consumer-tech",181,"2026-10-08T23:26:35.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Startups","startups",117,"2026-10-08T16:45:00.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":18},"Dev Tools","dev-tools",106,{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",66,"2026-10-09T04:46:11.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Gaming","gaming",58,"2026-10-08T20:08:45.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]