Unverified50% confidenceTradeoffExact time
System-level prompt injection is prevalent in commercial APIs for content moderation and safety alignment, and overly aggressive or poorly designed system prompts can theoretically interfere with a model's chain-of-thought performance.
1
Sources
50%
Confidence
Long-term
Relevance
8/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified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72% 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.71% similarUnverified商业化大模型服务会在用户可见对话之外附加系统提示，用于设定模型角色、行为规范、安全边界及输出风格，系统提示随每次API请求完整发送至模型69% similarUnverifiedContent passed when calling public APIs is often collected by service providers as training data for subsequent model iteration, which is why enterprise-level applications need to intercept sensitive information through a gateway.68% similarUnverifiedIn programming assistant scenarios, system prompts and code context typically consume a large number of tokens, resulting in an extremely high cache hit rate.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/828965API
curl https://kongchang.com/api/v1/knowledge/claims/828965MCP
get_claim(id=828965)