12000+ API Keys and Passwords Found in Public Datasets Used for LLM Training
It was discovered that about 12,000 live secrets, which enable successful authentication, are present in a dataset used to train large language models (LLMs).
The results demonstrate once more the serious security risk that hard-coded credentials present to both individuals and organizations. The issue is further exacerbated when LLMs recommend unsafe coding techniques to their users.
According to Truffle Security, it obtained an archive from December 2024 from Common Crawl, which keeps an open, free database of web crawl data. The enormous collection spans 18 years and includes more than 250 billion pages read more about 12,000+ API Keys and Passwords Found in Public Datasets Used for LLM Training.
Get up to date on the latest cybersecurity news and enhance your knowledge of cyb...