Unverified50% confidenceFactExact time
This three-level escalation mechanism was originally designed to combat child sexual exploitation content and terrorist content, and has now been migrated to threat detection scenarios.
1
Sources
50%
Confidence
Long-term
Relevance
10/10/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSince 2024, multimodal jailbreak attacks have begun to emerge, with attackers embedding malicious instructions in images or audio to bypass text-level safety detection.67% similarUnverifiedCommon prompt injection attack techniques include role hijacking, instruction override, context poisoning, and data exfiltration induction63% similarUnverifiedPrompt注入攻击是当前Agent安全领域最受关注的威胁之一62% similarUnverifiedBetween 2023 and 2024, security researchers discovered malicious models on the Hugging Face platform multiple times, with attackers executing attacks by embedding malicious code in model serialization files.62% similarUnverified攻击面通常分为三类：网络攻击面、软件攻击面和人员攻击面61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/993199API
curl https://kongchang.com/api/v1/knowledge/claims/993199MCP
get_claim(id=993199)