GitHub brings a purpose-built AI model to leaked secret detection

GitHub is rolling out a fine-tuned model for secret detection across its security tooling, the company announced on October 7, 2026. The model reads surrounding code to identify likely credentials, including passwords that have no recognizable token format, and GitHub says it does not generate code or prose.

Customers who already had AI-detected Password alerts have been automatically upgraded to the new model, according to GitHub. The company is also sharing plans to extend AI secret detection into push protection and into security reviews run through GitHub Copilot.

What is available now, and what is still in preview

AI-detected secrets in push protection is available in private preview, GitHub says. AI-based secret scanning through the GitHub Copilot /security-review command for the Copilot CLI and Copilot app is expected soon in private preview.

The model is also planned for AI-detected alerts in GitHub Enterprise Server 3.23 in public preview, included with an enterprise’s existing purchase of GitHub Secret Protection (GHSP) and GitHub Advanced Security (GHAS).

Alerts stay free, new checks consume AI Credits

GitHub says AI-detected secret alerts remain included in GHSP and GHAS at no additional charge. Existing AI-detected secret alert scans switch to the new model automatically at that same price, separate from the new credit-consuming checks.

The opt-in checks for push protection and the /security-review command will consume GitHub AI Credits. GitHub is publishing the planned billing model ahead of broader availability so administrators can review access and spending before enabling anything. AI Credit usage for these checks is expected to be introduced in the coming weeks.

For push protection, GitHub says a check can consume credits even if it does not block a push. Usage is billed to the organization that owns the repository, except in user-namespace repositories for enterprise-managed users, where it is attributed to the pusher and applies to that user’s allocated credits. Billing starts once an organization opts into the public preview and enables the feature.

The /security-review command reviews active changes for security vulnerabilities and returns prioritized findings with remediation suggestions. GitHub says the review is read-only, existing Copilot policies and billing apply, and no GHSP or GHAS license is needed for the new secret-classifier checks. Those checks will be off by default, and running /security-review will not turn them on.

Who qualifies

AI push protection requires paid GHSP or GHAS coverage and is available to customers on GitHub Enterprise Cloud or GitHub Teams; public, private and internal repositories can qualify. An administrator must enable it, subject to organization or enterprise policies. On GHEC with data residency on ghe.com, AI push protection is planned with paid GHSP or GHAS coverage.

Individual Copilot plans — Pro, Pro+, Max, Free and Student — are eligible for security review checks with the new model, subject to access controls and credit consumption. Copilot Business and Copilot Enterprise are eligible on supported platforms, subject to invitation and administrator policies; GitHub notes these subscriptions do not replace the GHSP license required for AI push protection.

AI push protection and the Copilot security review command are not part of the GHES 3.23 release.

Setting a spending cap

Organization and enterprise administrators can disable the new capabilities by policy and set budgets for AI Credit usage, and GitHub says opting in will not override those controls. To set a dedicated budget, open Billing and licensing, select Budgets and alerts, choose SKU-level budget, then Advanced Security as the product and Secret Protection AI Credits as the SKU.

GitHub warns that budget alerts alone do not stop usage — administrators who want a spending cap should configure “Stop usage when budget limit is reached” where available. Anyone already using AI push protection in private preview should disable it before the billing change takes effect if they do not want that usage, since continued use will consume AI Credits.

Sources

• Purpose-built model for leaked secret detection — GitHub Changelog