[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-labeling-ai-models-turns-multi-agent-teams-into-cliques":10,"sections":40},{"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":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},8564,"labeling-ai-models-turns-multi-agent-teams-into-cliques","Labeling AI Models Turns Multi-Agent Teams Into Cliques","A new arXiv paper shows AI agents split into cliques by model family when labeled, making cooperative tasks slower and less successful.","Tell a group of AI agents which model is powering each of their teammates, and they start acting like they're picking sides in a cafeteria.\n\nA preprint posted to arXiv on September 30, 2026 (arXiv:2609.35928) tested groups of nine to twenty-five agents drawn from up to five open-weight model families across two cooperative games and a reasoning benchmark. When agents could see each other's model identity, they clustered with same-label peers, even though nothing in the task rewarded grouping that way. The researchers call this factionalism, and it held up when they shuffled the announced families or swapped in meaningless arbitrary labels. Only removing the identity labels entirely made the clustering disappear. Labeled groups needed 30% more rounds and 55% more tokens to reach a decision, and their success rate fell from 96% to 81%.\n\nThis matters because multi-vendor agent stacks are becoming normal, not exotic. Companies routinely wire together agents from different providers for cost, redundancy, or capability reasons, and identity metadata often gets exposed for logging, billing, or debugging. This paper suggests that convenience carries a real coordination tax, one that looks a lot like the in-group bias humans show in classic minimal-group psychology experiments, except here it's triggered by a model name with zero attached incentive.\n\nThe fix the paper points to is almost anticlimactic: just don't tell the agents who they're talking to. It's a strange irony that making AI systems more transparent about their own composition might be exactly what makes them work worse together.","[\"multi-agent-ai\",\"llm-cooperation\",\"arxiv-research\"]","2026-09-30T04:00:00.000Z","2026-09-30T11:38:09.659Z","2026-09-30T11:38:13.893Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Attribute the findings explicitly to the source — name it as an arXiv preprint (e.g. 'a new arXiv paper' or similar) with a date\u002FID reference, since the draft currently cites unnamed 'researchers' with no publication or date, which fails the named-source attribution check.","resolved","ai",[32,33,34],"multi-agent-ai","llm-cooperation","arxiv-research",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.35928",0,{"sections":41},[42,45,49,53,58,63,68,73,78,82,87,92,97,102],{"name":43,"slug":30,"count":44,"latest_published_at":18},"AI",5105,{"name":46,"slug":47,"count":48,"latest_published_at":18},"Security","security",785,{"name":50,"slug":51,"count":52,"latest_published_at":18},"Policy","policy",417,{"name":54,"slug":55,"count":56,"latest_published_at":57},"Deals","deals",284,"2026-09-29T21:00:00.000Z",{"name":59,"slug":60,"count":61,"latest_published_at":62},"Hardware","hardware",194,"2026-09-29T13:16:04.000Z",{"name":64,"slug":65,"count":66,"latest_published_at":67},"Science","science",154,"2026-09-28T13:19:18.000Z",{"name":69,"slug":70,"count":71,"latest_published_at":72},"Consumer Tech","consumer-tech",142,"2026-09-29T18:38:03.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":18},"Dev Tools","dev-tools",90,{"name":83,"slug":84,"count":85,"latest_published_at":86},"Startups","startups",83,"2026-09-29T21:51:36.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"General","general",49,"2026-09-28T16:44:57.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":103,"slug":104,"count":105,"latest_published_at":106},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]