Unverified50% confidenceFactExact time
Adversarial examples refer to inputs that attackers deliberately construct to deceive model classification.
1
Sources
50%
Confidence
Long-term
Relevance
10/9/2026
First Seen
Sources
Related Entities
Related Claims
Verified提示注入是一种针对大语言模型的攻击手法，攻击者通过在模型输入数据中嵌入恶意指令劫持模型的正常行为78% similarUnverifiedAdversarial distillation attackers systematically send large volumes of requests to the target model, collecting input-output pairs—especially responses containing complete reasoning chains—to train their own models and replicate the target model's capabilities.75% similarUnverifiedBy injecting fabricated reasoning, it is possible to forge a model's thought process and induce it to output content it otherwise should not, constituting a form of jailbreak attack.75% similarVerifiedPrompt injection attacks fundamentally exploit the inherent flaw of large models in failing to separate instructions from data, and are divided into two categories: direct injection and indirect injection.71% similarUnverifiedIf a model states its plan during the reasoning process, the monitoring system has an opportunity to intercept it before the deceptive behavior causes actual harm.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/991250API
curl https://kongchang.com/api/v1/knowledge/claims/991250MCP
get_claim(id=991250)