[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-models-that-get-the-right-answer-for-wrong-reasons":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},4257,"ai-models-that-get-the-right-answer-for-wrong-reasons","AI Models That Get the Right Answer for Wrong Reasons","A new benchmark called NormWorlds-CF tests whether AI models actually understand ethical rules or just pattern-match to correct answers.","A research team has released NormWorlds-CF, a benchmark designed to catch language models that ace normative reasoning tests without understanding the underlying logic.\n\nThe system uses a deterministic solver — not another AI model — to verify answers, generate proof certificates, and flag when a model reaches the right verdict through faulty reasoning. The benchmark includes 270 rule-world families and 1,080 test pairs, and it stages diagnostics to separate surface accuracy from genuine comprehension. Training a model only on final answers produced perfect scores on answer tasks but zero on falsification tasks, exposing a gap that standard benchmarks would miss entirely. The researchers also introduced a new training method, MR-GRPO, that awards partial credit based on reasoning structure rather than just correct outputs.\n\nThis matters because most AI safety and alignment work assumes that a model getting the right answer is a reasonable proxy for a model that has internalized the right rule. NormWorlds-CF directly challenges that assumption, and does so with a solver that removes the circularity of using one AI to judge another. The finding that answer-only training scores zero on falsification is the kind of result that should make teams shipping rule-following agents nervous.\n\nOOD transfer — whether any of this holds on rule worlds the model has never seen — remains unsolved, which is the part that would need to work before any of this moves from benchmark to deployment.","[\"ai\",\"benchmarks\",\"alignment\",\"reasoning\"]","2026-07-07T04:00:00.000Z","2026-07-07T21:31:09.515Z","2026-07-07T21:31:12.411Z","published",null,[],"ai",[24,26,27,28],"benchmarks","alignment","reasoning",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.03957",0,{"sections":35},[36,40,45,50,55,60,65,70,75,79,84,88,93,98],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":18},"Dev Tools","dev-tools",59,{"name":80,"slug":81,"count":82,"latest_published_at":83},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":85,"slug":86,"count":82,"latest_published_at":87},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":99,"slug":100,"count":101,"latest_published_at":102},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]