Unverified85% confidenceFactTime unknown
2023年,安全研究者Johann Rehberger展示了如何通过在网页中嵌入不可见文本,让ChatGPT在用户不知情的情况下将对话内容泄露给外部服务器
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn 2023, researchers demonstrated hiding malicious instructions in invisible text areas of web pages, successfully hijacking a GPT-4-based browser agent into performing unauthorized operations.68% similarUnverified2024年安全研究员Johann Rehberger公开演示了针对Microsoft 365 Copilot的提示注入攻击,通过邮件正文嵌入人眼不可见的恶意指令,当用户让Copilot总结收件箱时执行隐藏指令,将敏感邮件内容通过超链接参数外泄到攻击者服务器68% similarUnverified2023年ChatGPT因Redis客户端库漏洞导致部分用户对话标题暴露给其他用户68% similarUnverified2023年安全研究员Riley Goodside与Johann Rehberger等人演示了通过网页内容、邮件正文劫持GPT-4插件和AutoGPT的完整攻击链66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/22020API
curl https://kongchang.com/api/v1/knowledge/claims/22020MCP
get_claim(id=22020)