📄 中文摘要
OpenAI面临泄密问题，过去18个月中，关于产品路线图、安全协议、财务状况和高管离职的未经授权的信息频繁出现在媒体上。为了解决这一问题，OpenAI开发了一种定制版本的ChatGPT作为监控工具。该工具通过将发布的新闻文章与内部Slack消息、电子邮件和文档库进行交叉引用，帮助安全团队识别潜在的泄密者。这一系统的目标是提高信息安全性，防止在关键时刻发生更多的泄密事件。
OpenAI开发了一种监控工具以捕捉泄密者。该工具是ChatGPT。
📄 English Summary
OpenAI Built a Surveillance Tool to Catch Leakers. The Tool Is ChatGPT.
OpenAI has been facing a significant leak problem, with unauthorized disclosures about product roadmaps, safety protocols, financials, and executive departures appearing in the press consistently over the past 18 months. In response, OpenAI has developed a custom version of ChatGPT as a surveillance tool to identify leakers. This tool works by feeding published news articles into a specialized instance of ChatGPT, which cross-references the articles with internal Slack messages, emails, and document repositories. The aim of this system is to enhance information security and prevent further leaks during critical moments.