PrivacyHigh
Alleged Inclusion of 12,000 Live API Keys in LLM Training Data Reportedly Poses Security Risks
- Occurred
- Feb 28, 2025
- Published
- Feb 28, 2025
- Source
- AI Incident Database
Summary
A dataset used to train large language models allegedly contained 12,000 live API keys and authentication credentials. Some of these were reportedly still active and allowed unauthorized access. Truffle Security found these secrets in a December 2024 Common Crawl archive, which spans 250 billion web pages. The affected credentials could have been exploited for unauthorized data access, service disruptions, financial fraud, and a variety of other malicious uses.
Developers
- Common Crawl
- OpenAI
- Microsoft
Deployers
- Microsoft
- OpenAI
- Common Crawl
- Microsoft Azure OpenAI Service
Harmed parties
- AWS
- Slack
- Mailchimp
- Microsoft
- Intel
- Huawei
- PayPal
- IBM
- Tencent
Tags
- AI security risks
- Data leaks
- Credential theft
- LLM training vulnerabilities
- Unauthorized access
- Common Crawl
- Cybersecurity breach
- AI-assisted hacking
- API key exposure
- AI data persistence
Classification
Category confidence: 90%
Severity confidence: 85%
Rationale: Unauthorized exposure of live API keys in training data poses significant security and privacy risks.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8