GitHub Copilot code review now affects both Copilot and Actions

GitHub announced a change that is easy to underestimate: starting June 1, 2026, Copilot code review in private repositories will consume not only AI Credits, but also GitHub Actions minutes. In other words, one pull request review now touches two different billing counters.

For a public repository, nothing changes: Actions minutes remain free there. But if your team lives in private repos, this is no longer just “another AI review.” It is now another item in your GitHub bill.

What actually changed

The rule is simple.

One more detail matters: GitHub says Copilot code review runs through GitHub Actions using GitHub-hosted runners. The default is a standard GitHub-hosted runner. But GitHub also supports larger GitHub-hosted runners and self-hosted runners, and those are billed differently.

So this is not only a Copilot question. It is a workflow and billing question at the same time.

Why this matters for a team

The usual mistake is to treat Copilot review as separate from CI. It is not.

Now a single PR review can affect:

For a small repo, the effect may be tiny. But if you have:

then the change may surface in your GitHub Actions usage, not in the Copilot UI.

There is also a useful nuance from GitHub’s docs: you do not need GitHub Actions enabled in your organization to use the agentic capabilities in code review. But if GitHub-hosted runners are disabled at the org level, the full agentic experience is not available and reviews fall back to a more limited mode. Self-hosted runners remain an option in that case.

What to check now

I would do this in order.

1. Find every private repo with Copilot code review enabled

Start with inventory, not billing. You need a list of the repos where review actually runs.

2. Identify the runner model used for review

There are three cases:

That is not a cosmetic detail. Larger runners cost more, while self-hosted runners do not consume GitHub Actions minutes in the usual way.

3. Capture the current Actions baseline

Look at how many minutes the team already burns per month. Without that baseline, you cannot tell whether the change adds 5% or 50%.

4. Check budgets and alerts

If nobody is watching Actions usage, the new cost can slip by quietly. A simple threshold alert is better than no signal at all.

5. Tell the team before the date arrives

Not everyone reads changelogs. Keep it short:

Anti-patterns

1. Thinking this is only a Copilot license change

It is not. It is also an Actions billing change.

2. Assuming public repo behavior applies to private repos

It does not. Public repositories keep free Actions minutes.

3. Ignoring the runner model

Standard, larger, and self-hosted runners are different economics. If you skip this check, your estimate will be wrong.

4. Waiting until June 1 to see what happens

At that point it is not an audit anymore. It is a postmortem.

5. Leaving review on “because it is there”

If the team is not ready to track this cost, lock the rules first and expand rollout later.

Short action plan before June 1

Main takeaway: this is not just a new AI feature. It is a new bill for a familiar workflow. If you do the audit before June 1, the surprise factor stays low.