https://blog.jqknono.com/zh-cn/tags/ai-%E5%AE%89%E5%85%A8/Recent content in AI 安全 on jqknono BlogsHugo -- gohugo.iozh-cnhttps://blog.jqknono.com (jqknono)https://blog.jqknono.com (jqknono)Mon, 17 Aug 2026 10:00:00 +0800https://blog.jqknono.com/zh-cn/blog/2026/08/17/model-relay-security/Mon, 17 Aug 2026 10:00:00 +0800https://blog.jqknono.com (jqknono)https://blog.jqknono.com/zh-cn/blog/2026/08/17/model-relay-security/用一组图讲清不可信模型中转（模型路由）服务的五大攻击面：信息窃取、数据劫持、资源劫持、内容篡改与供应链投毒，以及对应的防守措施。