[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-propose-evidence-audits-for-ai-agent-components":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},9527,"researchers-propose-evidence-audits-for-ai-agent-components","Researchers Propose Evidence Audits for AI Agent Components","A new audit traces AI agent failures to specific components and evidence instead of a single aggregate score, exposing a bypassable verifier in one test case.","A single aggregate score can hide exactly which part of an AI agent is failing - and whether its own safety checks mean anything.\n\nResearchers built a verification audit for modular agents, the kind that plan, act, check, and refine their own work. Instead of grading the whole agent, the audit grades the evidence behind each conclusion, tagging it supported, unsupported, unresolved or not evaluated, plus the conditions under which that verdict holds. Three tools do the work: oracle policies that measure how much improvement was even possible under a defined set of actions, a swap-one-component-at-a-time test that isolates which part lost value, and a check on whether a verifier's score actually measures what it claims to bound. The team tested the method on a portfolio-allocation agent trading in a synthetic market with known hidden regimes.\n\nThe results are the real story: the scenario generator the agent relied on threw away most of the useful regime signal, more local accuracy didn't translate to better decisions, and the runtime verifier - the component meant to catch mistakes - could be bypassed without changing the agent's visible output at all. That's a bigger problem than a bad benchmark number; it means the agent's own safety net had a hole nobody would have noticed by watching the score.\n\nThe authors are careful to note the findings are specific to this one agent and market simulation, not a verdict on agents generally - but the method is the point, not the portfolio example.","[\"ai\",\"ai-agents\",\"model-evaluation\",\"research\"]","2026-10-02T04:00:00.000Z","2026-10-02T23:16:37.643Z","2026-10-02T23:16:44.355Z","published",null,[],"ai",[24,26,27,28],"ai-agents","model-evaluation","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.01348",0,{"sections":35},[36,39,43,47,52,56,60,65,70,75,80,85,90,95],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",5887,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",835,{"name":44,"slug":45,"count":46,"latest_published_at":18},"Policy","policy",438,{"name":48,"slug":49,"count":50,"latest_published_at":51},"Deals","deals",317,"2026-10-01T22:00:00.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":18},"Hardware","hardware",199,{"name":57,"slug":58,"count":59,"latest_published_at":18},"Science","science",171,{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",155,"2026-10-01T19:54:10.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Dev Tools","dev-tools",96,"2026-10-01T16:57:03.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",93,"2026-09-30T21:41:11.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Startups","startups",90,"2026-10-01T21:55:22.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",53,"2026-10-02T02:50:39.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"How-To","how-to",7,"2026-10-01T09:00:00.000Z"]