[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-benchmark-finds-ai-crash-fixes-often-unsafe-in-real-code":10,"sections":33},{"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":28,"feedback":32,"feedback_at":22,"cost_usd":32,"total_tokens":32},9114,"new-benchmark-finds-ai-crash-fixes-often-unsafe-in-real-code","New Benchmark Finds AI Crash Fixes Often Unsafe in Real Code","A new benchmark shows AI can patch live software crashes up to 29% of the time, but a safety checker flags risky fixes along with many false alarms.","AI agents can now patch a crashing program while it is still running - and researchers just built the first real-world test of whether that is safe to trust.\n\nA new paper introduces HealBench, a benchmark of 265 runtime errors pulled from 18 real-world code repositories, each paired with a reference fix from the actual patched version. The researchers also built HealGuard, a safety layer that forces healing code into an analyzable subset of Python and uses static and dynamic taint analysis to track whether a fix's changes reach code a developer explicitly protected. Testing a dedicated healing method plus three general-purpose coding agents across three backbone LLMs, the best setup resumed execution in 38.11% of cases and actually passed the target test in 28.68%. Among the runs that passed, HealGuard flagged 17.4% as having changes that could reach protected operations - and in controlled tests it caught every unsafe case, though at a 68.42% false-positive rate.\n\nThat gap between the program kept running and the fix was actually correct is the real story here. Prior work on this kind of self-healing code tested toy competition problems, not messy production repositories, so a one-in-three success rate on real code is a meaningful, sobering baseline rather than a victory lap.\n\nA safety net that catches every real problem but cries wolf two-thirds of the time is still a safety net - just not one you can run unattended yet.","[\"ai\",\"dev-tools\",\"research\"]","2026-10-01T04:00:00.000Z","2026-10-01T20:42:05.884Z","2026-10-01T20:42:08.809Z","published",null,[],"ai",[24,26,27],"dev-tools","research",[29],{"name":30,"url":31},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.39086",0,{"sections":34},[35,38,42,46,51,56,60,65,69,73,78,83,88,93],{"name":36,"slug":24,"count":37,"latest_published_at":18},"AI",5572,{"name":39,"slug":40,"count":41,"latest_published_at":18},"Security","security",815,{"name":43,"slug":44,"count":45,"latest_published_at":18},"Policy","policy",430,{"name":47,"slug":48,"count":49,"latest_published_at":50},"Deals","deals",298,"2026-09-30T21:00:26.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Hardware","hardware",196,"2026-09-30T13:00:00.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":18},"Science","science",163,{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",149,"2026-09-30T22:57:11.000Z",{"name":66,"slug":26,"count":67,"latest_published_at":68},"Dev Tools",93,"2026-10-01T02:30:48.000Z",{"name":70,"slug":71,"count":67,"latest_published_at":72},"Software","software","2026-09-30T21:41:11.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Startups","startups",84,"2026-09-30T20:39:09.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":82},"Gaming","gaming",51,"2026-09-30T16:24:30.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":87},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]