[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-pair-humans-with-llms-to-audit-privacy-risk":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},5460,"researchers-pair-humans-with-llms-to-audit-privacy-risk","Researchers Pair Humans With LLMs to Audit Privacy Risk","Researchers propose pairing LLMs with human experts to assess privacy risks in CBDC-based welfare schemes, splitting document work from judgment calls.","A team of researchers has built a framework that pairs human experts with large language models to hunt for privacy risks in digital welfare systems.\n\nThe framework, called PRIAM, targets CBDC-based welfare schemes: government benefit programs run on central bank digital currency, which collect enormous amounts of beneficiary data. In the proposed process, an LLM first digs through documentation to draft data categories and privacy attributes. Human experts then check that work, separating what's actually documented from what the model merely inferred, flagging gaps, and sending unclear outputs back for revision. The researchers tested the approach on a data-characterization exercise using a CBDC welfare use case.\n\nPrivacy risk assessments for these systems are typically slow, manual slogs through stacks of technical and legal documents, work that's easy to under-resource and easy to get wrong. Handing the document-mining to an LLM while keeping judgment calls with humans is a plausible way to make these audits faster without pretending a model can replace expert scrutiny. That distinction matters especially for CBDC welfare programs, which combine centralized payment rails with vulnerable populations, a mix that makes surveillance and discrimination risks unusually high.\n\nThe paper is a proof of concept, not a deployed audit tool, so the real test is whether agencies actually adopt a process that still requires plenty of human hours.","[\"ai\",\"privacy\",\"cbdc\",\"llm\"]","2026-08-18T04:00:00.000Z","2026-08-18T21:09:33.572Z","2026-08-18T21:09:45.461Z","published",null,[],"ai",[24,26,27,28],"privacy","cbdc","llm",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.16461",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"]