**Scanning 7.6 Petabytes of HuggingFace Training Data for Secrets**

> We scanned every public dataset on Hugging Face for live secrets: 7.6 petabytes across 187 million files. We found 221,303 working credentials, including tokens that can push code into software millions of people install. AI agents leak secrets faster than you can fix them - schedule a demo with us at Black Hat 2026, Booth 5727 TRUFFLEHOG CUSTOMERS COMPANY RESOURCES LOG IN Contact Us AI agents leak secrets faster tha...

**来源信息**
- **来源**:Hacker News:AI 热帖
- **分类**:ai-products
- **发布时间**:2026-08-02 02:21(北京时间)
- **原文**:[打开原文](https://trufflesecurity.com/blog/scanning-7-6-petabytes-of-ai-training-data-for-secrets)