[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-audit-shows-ai-reasoning-text-isnt-always-the-real-reasoning":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},6696,"new-audit-shows-ai-reasoning-text-isnt-always-the-real-reasoning","New Audit Shows AI Reasoning Text Isn't Always the Real Reasoning","A causal audit of AI chain-of-thought reasoning finds the explanations often don't drive the actual computation, especially in large untuned models.","A new causal audit finds that AI models' chain-of-thought text often doesn't drive what the model is actually computing.\n\nResearchers built a metric called the CoT Mediation Index, which measures how much a model's output actually depends on its stated reasoning by patching the hidden states tied to that reasoning and comparing the resulting performance drop to a control patch. Testing multiple model families, including Phi, Qwen, and DialoGPT, across different scales, they found that reasoning's causal influence is usually concentrated in narrow layers rather than spread throughout the network. Some models scored near-zero on the index despite producing chain-of-thought text that looked perfectly plausible, meaning the explanation was cosmetic window dressing. Models specifically tuned for reasoning showed stronger, more structured dependence on their stated reasoning than larger models that hadn't been tuned that way, while separately, Mixture-of-Experts models showed a more distributed pattern consistent with how they route computation across experts.\n\nThe finding matters because chain-of-thought output is increasingly treated as a window into how a model reached an answer, including for safety monitoring and debugging. If the visible reasoning can be decoupled from the real computation, reading it tells you less than it appears to, and behavioral benchmarks alone can't catch the gap.\n\nChain-of-thought was sold as reasoning made visible; this audit is a reminder that fluent explanations and faithful ones are not the same claim.","[\"ai\",\"interpretability\",\"chain-of-thought\",\"ai-safety\"]","2026-09-17T04:00:00.000Z","2026-09-18T06:34:37.010Z","2026-09-18T06:34:48.924Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Restore the source's actual baseline for the reasoning-tuned-model finding — the paper says reasoning-tuned models show stronger, more structured mediation than larger untuned models, not versus MoE models as the current 'while Mixture-of-Experts models spread computation more diffusely' phrasing implies; state each comparison against its real baseline.","resolved","ai",[30,32,33,34],"interpretability","chain-of-thought","ai-safety",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.03994",0,{"sections":41},[42,46,50,55,60,64,68,73,78,82,87,92,97,102],{"name":43,"slug":30,"count":44,"latest_published_at":45},"AI",3853,"2026-09-17T08:27:09.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":18},"Security","security",648,{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",338,"2026-09-11T04:00:00.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":18},"Hardware","hardware",154,{"name":65,"slug":66,"count":67,"latest_published_at":18},"Science","science",114,{"name":69,"slug":70,"count":71,"latest_published_at":72},"Consumer Tech","consumer-tech",99,"2026-09-09T17:27:33.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Software","software",75,"2026-09-10T20:41:21.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":18},"Dev Tools","dev-tools",73,{"name":83,"slug":84,"count":85,"latest_published_at":86},"Startups","startups",55,"2026-09-09T23:14:29.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Gaming","gaming",43,"2026-09-10T12:18:06.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"General","general",41,"2026-09-08T01:57:23.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":103,"slug":104,"count":105,"latest_published_at":106},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]