As cited
Copy frozen at (site build).
cloud saas
GitHub adopts cheaper AI model to catch secrets developers leave in code
GitHub deployed a smaller, specialized credential classifier to detect secrets in code, replacing a large language model (LLM) pipeline. The new approach reduced scan time by 86% and model serving costs by over 98% while maintaining secret detection capability across push protection and post-push scanning stages.
Why it matters: Developers using GitHub Secret Scanning benefit from faster feedback and lower operational overhead, enabling quicker remediation of exposed credentials before they reach production.
- Source published
- First seen by Cybersecurity Tracker