Verified70% confidenceFactExact time
Prompt injection attacks can use carefully crafted user inputs to induce the model to ignore the constraints of the system prompt.
4
Sources
70%
Confidence
Long-term
Relevance
7/5/2026
First Seen
Sources
从零搭建AI Agent：用Python实现对话智能体核心功能
bilibili川府于瑾年7/2/2026
Related Claims
Unverified在多人协作线程中，提示注入（Prompt Injection）的攻击面扩大，任何有访问权限的用户都可能成为注入向量，恶意指令可能被伪装成背景补充或方向修正74% similarUnverifiedSystem prompts are a layer of instructions preset by vendors and invisible to users, injected into the model context before the user's conversation begins, used to shape the model's behavioral boundaries, response style, and safety policies.72% similarUnverified许多模型对System Prompt有特殊处理机制，其权重通常高于普通对话内容，将核心规则写入系统提示比放在普通用户消息中更能抵抗衰减71% similarVerified提示词注入（Prompt Injection）分为直接注入和间接注入两类，直接注入通过用户输入植入恶意指令，间接注入将恶意指令藏匿于AI可能读取的外部内容中69% similarUnverified自定义指令本质上是一种系统提示词（System Prompt）的持久化机制，将用户规则注入系统提示词层69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/112308API
curl https://kongchang.com/api/v1/knowledge/claims/112308MCP
get_claim(id=112308)