Unverified50% confidenceOpinionExact time
Indirect prompt injection infiltrates through channels such as tool return content and retrieved documents, posing a greater threat to agentic systems because agents trust tool output by default.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在具备工具调用能力的 Agentic 系统中，一次成功的提示注入可能导致数据泄露、恶意代码执行甚至供应链攻击78% similarUnverifiedAfter agents gain tool-calling and external system access capabilities, the attack surface expands significantly, facing risks such as prompt injection, unauthorized operations, and data leakage76% similarUnverifiedIntegrating content moderation does not equal security—if a deprecated old API is being called, the moderation will be as good as nonexistent.71% similarUnverified通过非官方平台获取的API密钥，其请求经中间服务器转发，对话内容、代码、提示词可能被中间方记录，存在敏感数据泄露风险70% similarVerifiedIndirect prompt injection attacks plant hidden instructions within normal business data to induce Agents into executing unauthorized operations, representing a security risk at the semantic level rather than the code level.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1020758API
curl https://kongchang.com/api/v1/knowledge/claims/1020758MCP
get_claim(id=1020758)