{"id":"seebug-paper-2fd8d17312bf","seq":1577357,"kind":"news","title":"当世界说谎：面向下游控制的潜在世界模型后门攻击","url":"https://paper.seebug.org/3520","permalink":"https://vulnpulse.app/items/seebug-paper-2fd8d17312bf","summary":"作者：Roberto Riaño, Gorka Abad, Stjepan Picek, Aitor Urbieta 原文链接：https://arxiv.org/pdf/2609.15781 摘要 预训练世界模型——一种学习得到的模拟器，它将观测编码为潜在状态并预测该状态在动作作用下如何演化——正开始像如今的预训练编码器和语言模型那样，被作为现成的动力学骨干网络复用于控制任务。我们表明，这种复用...","severity":null,"techs":[],"cves":[],"tags":[],"source":{"id":"seebug-paper","name":"Seebug Paper (Knownsec 404)"},"published_at":"2026-09-17T08:19:18.000Z","fetched_at":"2026-10-27T09:57:34.907Z"}