OpenAI、Anthropic、Google API漏洞泄露加密推理痕迹
"But marinade" and leaked passwords are what researchers found in ChatGPT's hidden reasoning
安全研究员挖出OpenAI、Anthropic、Google的API漏洞，能偷看模型隐藏推理，还泄露密码，得看看你的数据安不安全。
安全研究人员发现OpenAI、Anthropic和Google的API存在漏洞，可提取加密的推理痕迹并在模型间转移。
扫描公共会话时发现数十个密码和API密钥。推理摘要常隐藏模型实际行为。
研究还发现“But marinade”等异常内容。
"But marinade" and leaked passwords are what researchers found in ChatGPT's hidden reasoning
Security researchers found a vulnerability in the APIs of OpenAI, Anthropic, and Google that lets them extract encrypted reasoning traces and move them between models. A scan of public sessions turned up dozens of passwords and API keys. The traces also show that the reasoning summaries users see often hide what the models are actually doing. The article "But marinade" and leaked passwords are what researchers found in ChatGPT's hidden reasoning appeared first on The Decoder .
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Key entities and technical topics under continuous tracking include OpenAI, Anthropic, Google, API安全.
Source attribution: curated from Decoder (@Matthias Bastian). Cross-source corroboration: 11 independent sources have covered this development (including Yangyi, IT之家, arXiv cs.AI), confirming industry-wide signal strength.
