[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-agents-learn-to-catch-broken-rules-in-synthetic-data":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},5164,"ai-agents-learn-to-catch-broken-rules-in-synthetic-data","AI Agents Learn to Catch Broken Rules in Synthetic Data","A new workflow has AI agents hunt for hidden data rules, like matching totals, then automatically fix synthetic tables that break them.","A new research workflow teaches AI agents to hunt down the hidden rules in your data and enforce them on the fake stuff.\n\nResearchers describe a system where LLM agents discover three kinds of inter-column constraints in tabular data: equations (a total that must equal the sum of its parts), inequalities (a discount that can't exceed the price), and logical dependencies (if a field says \"cancelled,\" another field must be empty). The agents turn these into machine-checkable hypotheses, then a postprocessor plugs into any existing tabular data generator, checks its output against every retained constraint, and repairs violations without touching the generator itself. In evaluations, the full workflow caught more rule violations than simply asking an LLM to spot them in one shot, and the repair step produced zero measured violations for every constraint it kept, while mostly improving downstream utility and largely preserving the statistical shape of individual columns.\n\nSynthetic data is the workaround of choice when real records are scarce, sensitive, or locked behind privacy law - hospitals, banks, and fraud teams all lean on it to train models without exposing real customers. But data that looks statistically realistic can still be logically nonsensical, like a shipment that arrives before it ships, and models trained on that kind of noise inherit the nonsense. Separating \"does this look statistically right\" from \"does this actually make sense\" is a distinction most synthetic-data tools skip entirely.\n\nThe catch: the system can only enforce constraints it thinks to hypothesize in the first place, so it's a solid cleanup crew, not a guarantee the data is airtight.","[\"ai\",\"synthetic-data\",\"llm-agents\",\"data-generation\"]","2026-08-18T04:00:00.000Z","2026-08-18T07:48:00.389Z","2026-08-18T07:48:12.148Z","published",null,[],"ai",[24,26,27,28],"synthetic-data","llm-agents","data-generation",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.15109",0,{"sections":35},[36,40,44,49,54,59,64,69,74,78,83,88,93,98],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",3293,"2026-08-20T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":39},"Security","security",435,{"name":45,"slug":46,"count":47,"latest_published_at":48},"Policy","policy",210,"2026-08-19T09:32:27.000Z",{"name":50,"slug":51,"count":52,"latest_published_at":53},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":55,"slug":56,"count":57,"latest_published_at":58},"Hardware","hardware",140,"2026-08-19T18:25:42.000Z",{"name":60,"slug":61,"count":62,"latest_published_at":63},"Consumer Tech","consumer-tech",95,"2026-08-18T16:05:00.000Z",{"name":65,"slug":66,"count":67,"latest_published_at":68},"Science","science",90,"2026-08-19T18:41:02.000Z",{"name":70,"slug":71,"count":72,"latest_published_at":73},"Software","software",73,"2026-08-18T07:51:50.000Z",{"name":75,"slug":76,"count":77,"latest_published_at":18},"Dev Tools","dev-tools",69,{"name":79,"slug":80,"count":81,"latest_published_at":82},"Startups","startups",47,"2026-08-19T19:13:46.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":87},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"General","general",33,"2026-08-18T22:18:13.000Z",{"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"]