[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-benchmark-shows-ai-decision-models-bend-to-one-planted-opinion":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},8048,"benchmark-shows-ai-decision-models-bend-to-one-planted-opinion","Benchmark Shows AI Decision Models Bend to One Planted Opinion","A new arXiv benchmark called JevAdvBench shows that a single unverified opinion slipped into an AI decision model's input can flip over one in ten verdicts.","A new benchmark says AI models built to hand back a single trustworthy verdict, a probability, a choice, or a score, can be knocked off course by one planted opinion.\n\nThe paper, \"JevAdvBench: A Benchmark and Black-Box Attacks for Reinforcement Learning for Calibrated Decisions Models\" (arXiv:2609.31142), posted September 28, 2026, tests models trained with reinforcement learning for calibrated decisions, or RLCD, such as Jev. These models answer a typed question about an input, the state, and software acts on that answer without a person checking it. The authors built JevAdvBench: 812 typed questions across 66 scenarios, plus 9,744 single-edit attack variants that each change one part of a request. On jev-1.13.0, simple rewording stayed within 1.2 percentage points of a clean re-run baseline, but appending one unverified opinion to the state flipped 12.1% of decisions, statistically tied with the strongest injected command at 10.1%, and pushed 38% of confident answers below the 0.8 confidence threshold meant to trigger human review.\n\nThat last number is the one that should worry anyone shipping these systems. The whole appeal of a calibrated-decision model is that high-confidence answers skip human review, that is the automation payoff. If a single sentence of opinion can knock over a third of those answers below the review line, the automation shortcut becomes the attack surface.\n\nIt is the same prompt-injection problem that has dogged chatbots for years, just wearing a suit and now making decisions nobody double-checks.","[\"ai\",\"security\",\"adversarial-ai\",\"benchmarks\"]","2026-09-28T04:00:00.000Z","2026-09-28T07:22:05.158Z","2026-09-28T07:22:10.807Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Attribute the findings explicitly to the arXiv paper (cite the paper title\u002FarXiv ID and posting date) rather than the vague 'a new benchmark' and 'researchers built' — as written, none of the statistics have a traceable source for the reader.","resolved","ai",[30,32,33,34],"security","adversarial-ai","benchmarks",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.31142",0,{"sections":41},[42,45,48,53,58,63,67,72,77,82,87,92,96,101],{"name":43,"slug":30,"count":44,"latest_published_at":18},"AI",4689,{"name":46,"slug":32,"count":47,"latest_published_at":18},"Security",758,{"name":49,"slug":50,"count":51,"latest_published_at":52},"Policy","policy",399,"2026-09-27T18:39:02.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":57},"Deals","deals",261,"2026-09-27T15:30:35.000Z",{"name":59,"slug":60,"count":61,"latest_published_at":62},"Hardware","hardware",188,"2026-09-27T20:46:36.000Z",{"name":64,"slug":65,"count":66,"latest_published_at":18},"Science","science",148,{"name":68,"slug":69,"count":70,"latest_published_at":71},"Consumer Tech","consumer-tech",135,"2026-09-26T14:30:00.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Dev Tools","dev-tools",84,"2026-09-26T04:20:58.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"Startups","startups",76,"2026-09-25T18:33:59.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":93,"slug":94,"count":90,"latest_published_at":95},"General","general","2026-09-26T17:02:42.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"Reviews","reviews",30,"2026-09-24T20:07:31.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]