[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-framework-exposes-wild-swings-in-ais-hidden-water-use":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},8261,"new-framework-exposes-wild-swings-in-ais-hidden-water-use","New Framework Exposes Wild Swings in AI's Hidden Water Use","SustainAI tracked AI water use across data centers and found an 11-fold gap, with most tested responses turning out wrong anyway.","A new research framework wants AI systems to be judged by how much water they burn per answer, not just how many they get right.\n\nSustainAI, described in a new arXiv paper, is a closed-loop framework that pairs real-time water metering with a hallucination-aware penalty model and a routing algorithm that steers requests based on regional water stress. Researchers tested it on small language models tasked with extracting health misinformation, and found an 11-fold gap in water footprint depending on which data center handled the request, from 0.0477 mL to 0.5360 mL per inference. Across 1,335 inference runs, the system consumed roughly 399 mL of water in total. Only 240 of those runs produced a correct answer.\n\nThat ratio is the real headline: most of the water went toward wrong answers, not right ones. Carbon and energy use already get climate scrutiny, but water, especially in regions under drought stress, has stayed a footnote in AI's environmental accounting. Tying accuracy to resource cost, rather than treating them separately, gives cloud providers and enterprise buyers a harder number to route around.\n\nIt's a small-scale academic test, not a production audit of a hyperscaler, but it hints at the kind of scorecard AI vendors will eventually face as water scrutiny catches up with carbon scrutiny already applied to tech supply chains.","[\"ai\",\"sustainability\",\"water-usage\",\"data-centers\"]","2026-09-28T04:00:00.000Z","2026-09-28T20:15:39.194Z","2026-09-28T20:15:45.650Z","published",null,[],"ai",[24,26,27,28],"sustainability","water-usage","data-centers",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.30747",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,90,94,99],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",4881,"2026-09-28T16:14:54.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",765,"2026-09-28T15:35:23.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",401,"2026-09-28T15:14:48.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",268,"2026-09-28T14:00:00.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",191,"2026-09-28T15:45:00.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Science","science",154,"2026-09-28T13:19:18.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Consumer Tech","consumer-tech",139,"2026-09-28T17:09:47.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",85,"2026-09-28T13:00:00.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Startups","startups",78,"2026-09-28T15:29:21.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":91,"slug":92,"count":88,"latest_published_at":93},"General","general","2026-09-26T17:02:42.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]