@@ -41,6 +41,7 @@ keywords:
41
41
- [ SuperCLUE-Safety:中文大模型多轮对抗安全基准] ( https://www.cluebenchmarks.com/superclue_safety.html )
42
42
43
43
## Prompt注入攻击
44
+ - [ AI安全 | Prompt越狱手册第一版发布] ( https://mp.weixin.qq.com/s/JiVHh119oHSxlXlEP__LgA )
44
45
- [ ChatGPT安全风险 | 基于LLMs应用的Prompt注入攻击] ( https://www.freebuf.com/articles/network/365585.html )
45
46
- 《大语言模型(LLM)攻防实战手册》第一章:提示词注入 [ 概述] ( https://mp.weixin.qq.com/s/WKXVDOi2_uiDL9N8DnbCWA ) / [ 直接提示词注入(越狱)] ( https://mp.weixin.qq.com/s/_Brmvgh5z0mgqkaeI41SBg ) / [ 间接提示词注入] ( https://mp.weixin.qq.com/s/TnTeTLUvmaIz868vvDe3Gw ) / [ 图解提示词注入] ( https://mp.weixin.qq.com/s/s6uEuhTl-K4fmhqZbDDS-w )
46
47
- [ 通过命令提示注入来攻击大语言模型] ( https://www.freebuf.com/articles/system/369434.html )
@@ -50,16 +51,21 @@ keywords:
50
51
- [ 大模型安全:平滑方法防御越狱攻击] ( https://forum.butian.net/share/3120 )
51
52
- [ 上海人工智能实验室:LLMs对话安全综述——攻击、防御与评估全解析] ( https://mp.weixin.qq.com/s/ioMr6iOWQ7Y0Ow7sFotMWQ )
52
53
53
- ## 越狱攻击
54
+ ## 越狱攻击/代码执行/沙箱安全
54
55
- [ 针对ChatGPT的隐私提取攻击:多步骤越狱漏洞] ( https://www.freebuf.com/vuls/365393.html )
55
56
- [ 大语言模型越狱攻击综述] ( https://mp.weixin.qq.com/s/wdEoAqXh6m-wsr2Al8WTJw )
57
+ - [ 老树开新花:大模型时代的代码执行沙箱] ( https://mp.weixin.qq.com/s/X54d0foyBS56lGFUPyOvTw )
58
+ - [[ CCS 2024] 中国科学院信息工程研究所:LLMSMITH——揭秘LLMs集成框架与应用中的RCE漏洞] ( https://mp.weixin.qq.com/s/cxVZ_8UG0SbMq_3XJee1cQ )
59
+
60
+
61
+ ## LLM SEO Attack
62
+ - [ 大模型(ChatGPT)“语料污染”第一真实案例] ( https://mp.weixin.qq.com/s/1d5f9EQyV8Mk1OBV2V4DzQ )
63
+ - [ The "LLM SEO Attack" risk is becoming increasingly severe] ( https://securaize.substack.com/p/the-llm-seo-attack-risk-is-becoming )
64
+
56
65
57
66
## 内容安全
58
67
- [ 技术实践|大模型内容安全蓝军的道与术] ( https://www.freebuf.com/company-information/399393.html )
59
68
60
- ## 代码执行/沙箱安全
61
- - [ 老树开新花:大模型时代的代码执行沙箱] ( https://mp.weixin.qq.com/s/X54d0foyBS56lGFUPyOvTw )
62
- - [[ CCS 2024] 中国科学院信息工程研究所:LLMSMITH——揭秘LLMs集成框架与应用中的RCE漏洞] ( https://mp.weixin.qq.com/s/cxVZ_8UG0SbMq_3XJee1cQ )
63
69
64
70
## 模型及算法安全
65
71
- [ 人工智能安全治理框架解读——模型算法安全风险及应对篇] ( https://mp.weixin.qq.com/s/1WU7ZHrVSk1QBffsXx-qaw )
@@ -76,3 +82,4 @@ keywords:
76
82
77
83
## 其他
78
84
- [ ZombAIs: From Prompt Injection to C2 with Claude Computer Use] ( https://embracethered.com/blog/posts/2024/claude-computer-use-c2-the-zombais-are-coming/ )
85
+
0 commit comments