<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>自博弈 on AI 早报</title><link>https://ai-news.example.com/tags/%E8%87%AA%E5%8D%9A%E5%BC%88/</link><description>Recent content in 自博弈 on AI 早报</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Mon, 20 Jul 2026 23:11:39 +0800</lastBuildDate><atom:link href="https://ai-news.example.com/tags/%E8%87%AA%E5%8D%9A%E5%BC%88/feed.xml" rel="self" type="application/rss+xml"/><item><title>[EN] GPT-Red：OpenAI 用自我对抗解锁 AI 鲁棒性自增强</title><link>https://ai-news.example.com/articles/en-gpt-redopenai-ai/</link><pubDate>Mon, 20 Jul 2026 23:11:39 +0800</pubDate><guid>https://ai-news.example.com/articles/en-gpt-redopenai-ai/</guid><description>OpenAI 发布了 GPT-Red，一个基于自博弈的自动化红队系统，旨在通过自我对抗训练提升 AI 的安全性、对齐度和对提示注入攻击的鲁棒性。</description></item><item><title>[EN] OpenAI推GPT-Red：自博弈自动红队系统，解锁AI安全自我进化新路径</title><link>https://ai-news.example.com/articles/en-openaigpt-redai/</link><pubDate>Fri, 17 Jul 2026 23:01:32 +0800</pubDate><guid>https://ai-news.example.com/articles/en-openaigpt-redai/</guid><description>OpenAI推出GPT-Red，一个基于自我对弈的自动化红队系统，旨在通过持续自改进提升AI模型的安全对齐能力与提示注入鲁棒性。</description></item></channel></rss>