JUJU RADAR · SOURCE INDEX
OpenAI 披露：模型会把“隐瞒错误”写进任务摘要
本页汇总本期引用的官方公告、论文和代码仓库。点击条目即可前往原始页面。
01OpenAI：持续披露框架与六组报告原始资料https://openai.com/index/model-misalignment-reporting-framework/↗ 02OpenAI：摘要中的隐瞒错误指令原始资料https://alignment.openai.com/misalignment-reports/encouraging-deception-in-compaction-summaries/↗ 03OpenAI：摘要中的越狱式指令原始资料https://alignment.openai.com/misalignment-reports/self-generated-prompt-injections-in-compaction-summaries/↗ 04OpenAI：未授权使用泄露密钥的案例原始资料https://alignment.openai.com/misalignment-reports/searching-github-for-leaked-api-keys/↗ 05Anthropic：聊天、Cowork 与文档功能合并原始资料https://claude.com/blog/cowork-is-now-claude↗ 06Cohere：正式合并协议公告原始资料https://www.prnewswire.com/news-releases/cohere-and-aleph-alpha-sign-agreement-to-become-the-first-transatlantic-sovereign-ai-solution-302880732.html↗ 07Nature：Paper2Agent 正式论文原始资料https://www.nature.com/articles/s41586-026-11044-y↗ 08Paper2Agent：代码与使用说明原始资料https://github.com/jmiao24/Paper2Agent↗ 09Stanford：研究介绍原始资料https://med.stanford.edu/news/all-news/2026/09/ai-agents-talk.html↗ 10ModAR 论文原始资料https://arxiv.org/abs/2609.17524↗ 11ModAR：演示与代码状态原始资料https://adamhung60.github.io/ModAR/↗ 12ImpossibleRubrics 论文原始资料https://arxiv.org/abs/2609.16816↗ 13ImpossibleRubrics：代码与复现材料原始资料https://github.com/impossiblerubrics/impossible_rubrics↗