Unverified50% confidenceEventExact time
In 2023, researchers demonstrated that simply writing an instruction in small white text on a web page to forward a user's email to a certain address would cause an LLM Agent browsing that page to automatically trigger the outbound operation.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
Unverified2023年已有研究者演示了通过在网页中嵌入白色文字指令，成功劫持支持网络访问的ChatGPT插件72% similarUnverifiedSome researchers have received official emails from NeurIPS 2026 regarding the new Automatic Reference Checker.66% similarUnverified2023年研究者已经演示了通过间接注入控制Bing Chat代理行为的概念验证攻击63% similarUnverifiedA researcher posted on Reddit's machine learning community, describing their experience of receiving a desk rejection during the AAAI-27 submission process.63% similarUnverifiedIn 2024, security researcher Johann Rehberger demonstrated an indirect prompt injection attack that successfully hijacked the output of multiple commercial RAG systems by embedding instruction text in public web pages that was invisible to humans but readable by AI63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/943795API
curl https://kongchang.com/api/v1/knowledge/claims/943795MCP
get_claim(id=943795)