GitHub Copilot code review can now approve pull requests in public preview when administrators explicitly enable the feature.
That is a meaningful shift in the role of coding assistants. Copilot review is no longer only an advisory layer that comments on code. In configured repositories, it can participate in the merge-control system by submitting an approval that counts toward required approvals.
GitHub says the feature is available to Copilot Pro, Pro+, Max, Business, and Enterprise plans. It is off by default.
Approval assessments come first
Every Copilot code review now includes an approval assessment in its overview comment. That assessment tells maintainers whether Copilot considers the pull request ready to approve.
The assessment alone does not satisfy merge requirements. It is a visible signal for the human reviewer or maintainer.
The bigger change appears when admins turn on approval submission. GitHub says Copilot can then submit an approval that counts toward a repository’s required-approval rule. If new commits are pushed after Copilot approves, the approval is dismissed like a human review, and teams can request a new Copilot review.
That behavior makes Copilot more similar to a controlled reviewer identity inside the development workflow.
Admin controls decide where it applies
GitHub says admins can control the feature at enterprise, organization, and repository levels.
At the enterprise level, admins can keep approvals off globally or let organizations decide. At the organization level, admins can turn approvals on, let repository admins decide, enable the feature for specific repositories, or turn it off. At the repository level, admins can turn approvals on or off and choose which file paths Copilot is allowed to approve.
The path control is important. Most organizations should not treat all files as equal. A documentation-only change, generated file, frontend copy tweak, infrastructure policy file, payment flow, authentication code, and cryptography change do not carry the same review risk.
If Copilot approvals are enabled, repository owners need a path policy that reflects that risk.
This lands during a broader Copilot governance push
The approval update follows several GitHub Copilot governance changes: global model policies, billing controls, model access updates, code-review effort levels, and new model availability such as Claude Fable 5.1 in Copilot.
Together, those changes point to a more administrative version of coding AI. The question is no longer simply whether developers like the assistant. It is how an organization governs model access, spend, retention, review authority, and agent behavior across all the places code moves.
Copilot approvals are especially sensitive because they touch the boundary between advice and authorization. A comment can be ignored. An approval can unblock a merge.
