Unverified50% confidenceFactExact time
In 2023, a Carnegie Mellon University research team discovered a gradient-based automated jailbreak method (the GCG attack), which can bypass the safety protections of almost all mainstream models by appending seemingly meaningless suffix strings to prompts.
1
Sources
50%
Confidence
Long-term
Relevance
8/16/2026
First Seen
Sources
AI智能体自发形成康德伦理？技术真相与拟人化陷阱
Related Claims
Unverified2024年以来，研究者发现了对抗性后缀攻击（Adversarial Suffix），即在正常提示后附加一串看似无意义的token序列就能让模型忽略安全对齐71% similarUnverified2024年研究揭示了多跳间接注入攻击，攻击者先污染模型记忆或工具调用状态，再通过后续交互完成数据外泄，使基于单次请求的检测系统失效65% similarUnverified2023年多项研究证明，即使经过RLHF对齐的模型也难以完全抵抗精心设计的间接提示词注入64% similarUnverified2023年多项学术研究证明，即使经过严格安全对齐训练的模型，当恶意指令以自然语言文档形式出现在上下文窗口中时，其遵循率仍显著高于直接注入场景63% similarVerifiedMultiple academic studies in 2024 indicate that current mainstream LLMs still have limited resistance to carefully crafted prompt injection attacks.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/756999API
curl https://kongchang.com/api/v1/knowledge/claims/756999MCP
get_claim(id=756999)