The 8 October change
GitHub’s 8 October announcement gives organization owners a choice to charge eligible licensed members’ code reviews to the organization instead of the member’s entitlement. Member billing remains the default. Organization billing requires paid AI-credit usage to be enabled; an explicit budget is optional.
The announcement also adds a control over reviews requested with externally supplied Copilot licenses, such as a personal license or one supplied by another organization. Billing ownership and permission to request a review are distinct controls.
A review has more than one cost component
The current code-review documentation separates AI credits for the model interaction from GitHub Actions usage for agentic context gathering and tool work. Changing the billing source does not itself grant code-review access. Automatic reviews configured through repository or organization rulesets also have behavior distinct from personal review settings.
GitHub publishes estimated credit-consumption ranges, with variation by review effort and workload; those estimates exclude Actions minutes. We have not converted them into a fixed cost per pull request or a promised monthly saving.
Map the payer before enabling paid usage
A small billing worksheet can expose the cases that a single seat-price comparison misses. Start with a hypothetical team and list representative review requests, without running or paying for them:
- A member requests a review. Record the license issuer, chosen billing source and applicable usage budget. Specify what should happen when that allowance is exhausted.
- An outside collaborator requests a review. Record repository access separately from the license they bring. Note which authorization rule is intended to apply, and verify the resulting UI or API behavior in an approved trial.
- A rule requests a review automatically. Write down the rule’s scope and trigger. Include repeated reviews after new commits when estimating how often a request might occur.
- The review gathers project context. Identify which runner configuration would be used and where its consumption can be inspected. Keep this line separate from model-credit consumption.
For each row, name the owner who can verify the billing report and the person allowed to approve a policy change. If a later trial is authorized, retain the request, attribution and usage record together. That is more useful evidence than multiplying a guessed review price by the number of developers.
What this update does not establish
This report covers source-documented controls, not an observed invoice. No paid-usage policy, budget or access setting was changed, and no billable review was requested. The Copilot profile provides the wider product record; the sandboxing update addresses local tool execution rather than review billing.
Sources and scope
Official sources read 10 October 2026. Facts above are attributed to the provider; the evaluation worksheets are proposed checks. This article does not report a native product test or assign a tool score.