Unverified50% confidenceFactExact time
ORA attack statements often pass source-support checks and are difficult to identify as malicious input.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
Unverified针对特定触发词的后门攻击可让模型在大多数情况下表现正常，仅在遇到特定输入时才输出预设错误信息，使常规评测基准难以捕捉64% similarUnverifiedIndirect prompt injection specifically refers to attackers carrying out attacks through external content that the model actively retrieves, making it more dangerous.62% similarUnverifiedTo counter jailbreak attacks, systems maintain high vigilance toward certain domain-specific word combinations, causing legitimate inter-institutional sample transfer requests and malicious queries to appear highly similar at the feature vector level.61% similarUnverifiedIn jailbreak attack scenarios, scoring an early prefix alone may not accurately predict its attack effectiveness when combined with a suffix, so pairwise evaluation is more reliable than evaluating prefixes individually.60% similarUnverifiedAPI field brute-forcing refers to exhaustively trying large numbers of automated attempts to enumerate possible parameter names, field values, or endpoint paths in order to discover undisclosed functionality or obtain restricted data.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1074747API
curl https://kongchang.com/api/v1/knowledge/claims/1074747MCP
get_claim(id=1074747)