Unverified50% confidenceCautionExact time
Models may generate commands that are semantically correct but behaviorally dangerous due to contextual misunderstanding—for example, expanding rm -rf ./tmp into rm -rf /, or omitting a WHERE clause in a SQL UPDATE statement.
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedPrompt injection attempts to override the system prompt and induce the model to perform unauthorized actions by embedding malicious instructions into external content processed by the model, and is logically analogous to SQL injection.69% similarUnverifiedIn the revision stage, LLMs can point out logical jumps, redundant sentences, or check whether arguments lack supporting evidence.68% similarUnverified提示词与模型版本存在深度绑定关系，不同版本模型在RLHF训练数据、指令微调策略上的差异会导致相同提示词产生不同输出，这种现象被称为「提示词漂移」66% similarUnverifiedLLM在对话中依赖上下文动态推理能临场识别异常，但生成的代码脚本固化后逻辑分支是静态的，无法感知未见过的数据模式，这被称为分布偏移（Distribution Shift）66% similarVerifiedThe output of LLMs is inherently non-deterministic; even with the same Prompt as input, the model may return results with different formats, content, or even logic at different times.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/985742API
curl https://kongchang.com/api/v1/knowledge/claims/985742MCP
get_claim(id=985742)