[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-fraud-detection-model-says-how-sure-it-really-is":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},5400,"new-fraud-detection-model-says-how-sure-it-really-is","New Fraud Detection Model Says How Sure It Really Is","ProtoCP, a new graph-based fraud detection model, gives statistically guaranteed confidence scores that are tighter and more useful than rival systems.","A new academic framework called ProtoCP promises fraud detection systems that know when they are guessing.\n\nResearchers describe ProtoCP, a conformal prediction framework for spotting fraud in temporal transaction graphs, networks of who paid whom and when. Conformal prediction is a statistical technique that wraps a model's raw score in a mathematically guaranteed confidence interval, rather than a single yes or no verdict. The team found that off-the-shelf conformal methods struggle on fraud data because genuine fraud cases sit in mostly-benign neighborhoods that dilute the calibration signal, and because so few transactions are actually labeled fraud that class-specific thresholds end up wildly conservative. ProtoCP addresses this with learned prototypes that filter out benign noise during calibration and a neighborhood-relative scoring method, smoothed over time, to handle shifting fraud patterns. Tested on four public benchmarks, YelpChi, S-FFSD, FTFD, and BankSim, it hit its target coverage guarantees while producing smaller, more decisive prediction sets than existing baselines.\n\nFraud teams do not just want a fraud or not-fraud label; they want to know how much to trust it, because a false positive can freeze a legitimate customer's account and a false negative can let money walk out the door. A system that can say a flag is backed by a formal confidence guarantee, rather than an opaque probability score, gives risk teams something they can actually build policy around, especially in the messy, imbalanced data real fraud systems produce.\n\nThe code is open-source on GitHub, so this reads more like a contribution for engineers building fraud pipelines than a product any bank will ship tomorrow. The real test will be whether it holds up on live, non-benchmark data, where fraud patterns are less cooperative than in these four curated datasets.","[\"fraud-detection\",\"conformal-prediction\",\"graph-neural-networks\",\"ai-research\"]","2026-08-18T04:00:00.000Z","2026-08-18T18:40:34.250Z","2026-08-18T18:40:47.945Z","published",null,[],"ai",[26,27,28,29],"fraud-detection","conformal-prediction","graph-neural-networks","ai-research",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.15768",0,{"sections":36},[37,41,45,50,55,60,65,70,75,79,84,89,94,99],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",3293,"2026-08-20T04:00:00.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":40},"Security","security",435,{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",210,"2026-08-19T09:32:27.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",140,"2026-08-19T18:25:42.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",95,"2026-08-18T16:05:00.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Science","science",90,"2026-08-19T18:41:02.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",73,"2026-08-18T07:51:50.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":18},"Dev Tools","dev-tools",69,{"name":80,"slug":81,"count":82,"latest_published_at":83},"Startups","startups",47,"2026-08-19T19:13:46.000Z",{"name":85,"slug":86,"count":87,"latest_published_at":88},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",33,"2026-08-18T22:18:13.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]