GitHub Enterprise Server 3.23 (release candidate) adds local AI secret detection
GitHub has released the GHES 3.23 release candidate with a new AI secret detection feature. According to the company, it recognizes unstructured secret data directly on the appliance without a GPU, a hosted LLM, or an internet connection, including in air-gapped environments. The feature is in public preview.
GitHub has released a release candidate of GitHub Enterprise Server (GHES) 3.23, which among other things brings the AI secret detection feature. According to GitHub, this is a separate, fine-tuned model that runs directly on the appliance and can recognize unstructured secrets in code without needing a GPU, a hosted LLM, or any outbound network connection. As a result, it also works in air-gapped environments, where the customer has no external internet connection.
The feature is part of GitHub Secret Protection and in this release candidate is available as a public preview, meaning, according to the company, it is still subject to change. The GHES 3.23 release candidate is now available for download for customers who want to try the new features before general availability and provide feedback to GitHub Support.
The source does not provide further technical details about detection accuracy, supported secret formats, or the timeline for general availability after the release candidate.
Why it matters
For organizations running GHES in isolated or regulated environments (banking, government, defense), it is crucial that detection of sensitive data such as passwords or keys in code does not go through an external cloud AI service but happens locally on the appliance. This reduces the risk that the analyzed code or discovered secrets could leave the customer's network, and enables deployment even where cloud AI tools are prohibited for security or compliance reasons.
Relevant practical impact
What this means
For a business
Companies using GitHub Secret Protection on GHES can now (in preview) enable detection of unstructured secrets without relying on external GPU infrastructure or a cloud LLM, which makes deployment in air-gapped and regulated environments easier and reduces the risk of source code leakage during scanning.
Risks and complianceCheck the original
Event sources
clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.