[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-find-a-smarter-way-to-aim-adversarial-noise":10,"sections":35},{"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":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},9676,"researchers-find-a-smarter-way-to-aim-adversarial-noise","Researchers Find a Smarter Way to Aim Adversarial Noise","A new method redirects a fixed adversarial noise budget toward a model's most sensitive points, boosting attack success without spending more.","A new adversarial attack method gets more mileage from the same perturbation budget by targeting it at the parts of an image a model actually cares about.\n\nResearchers describe an importance-guided allocation mechanism for adversarial attacks constrained by a shared l1 budget, where total perturbation is capped and no single pixel can be pushed too far. The method uses a fixed clean-gradient prior, essentially a map of which input regions the model is most sensitive to, and steers the attack's limited budget toward those spots instead of spreading it evenly. A centered allocation objective then rewards perturbation at above-average-importance locations and penalizes wasted effort elsewhere, redistributing the existing budget rather than requesting more of it. Tested across ten robust model-dataset pairs, the approach beat standard APGD- and PMA-based attacks by 2.52 to 17.70 percentage points in attack success rate, without increasing total perturbation.\n\nThis matters because it reframes a core assumption behind adversarial robustness testing. APGD and PMA are standard tools for stress-testing defenses, and this result suggests even well-established attack baselines are not using their allotted noise optimally, meaning current robustness numbers could be softer than they look once attackers simply spend the same budget more intelligently.\n\nIn other words: the lock wasn't picked with a bigger key, just a better-cut one.","[\"adversarial-attacks\",\"ai-security\",\"robustness-research\",\"arxiv\"]","2026-10-02T04:00:00.000Z","2026-10-03T05:43:24.898Z","2026-10-03T05:43:29.038Z","published",null,[],"security",[26,27,28,29],"adversarial-attacks","ai-security","robustness-research","arxiv",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.00861",0,{"sections":36},[37,41,44,48,53,57,61,66,71,76,81,86,91,96],{"name":38,"slug":39,"count":40,"latest_published_at":18},"AI","ai",5977,{"name":42,"slug":24,"count":43,"latest_published_at":18},"Security",842,{"name":45,"slug":46,"count":47,"latest_published_at":18},"Policy","policy",438,{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",317,"2026-10-01T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",199,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",173,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",155,"2026-10-01T19:54:10.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Dev Tools","dev-tools",96,"2026-10-01T16:57:03.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",93,"2026-09-30T21:41:11.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Startups","startups",90,"2026-10-01T21:55:22.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Gaming","gaming",53,"2026-10-02T02:50:39.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",7,"2026-10-01T09:00:00.000Z"]