GitHub previews Copilot billing, unified-agent, and code-review policy changes

GitHub's September schedule reopens paid Business and Enterprise signups, unifies Copilot agent surfaces, changes chat retention, and makes Balanced the default review effort.

GitHub published a set of Copilot policy and billing changes on August 28, 2026, affecting enterprise payments, agent entry points, data retention, and the default code-review setting. This is not one feature launch; it is a move toward a shared Copilot agent experience. The announcement gives several effective dates, so administrators should compare each one with their own billing cycle and organization policy.

The first change starts September 1: GitHub will reopen card and PayPal signups for new Copilot Business and Enterprise customers. New seats must be paid for before access is granted, and from the next billing cycle all assigned seats will be billed upfront. Existing card and PayPal customers are expected to see the change from October 1, while overage usage may require additional payment. GitHub says prices are unchanged, but removing a seat will not create a prorated refund; spend controls and usage tracking remain available.

The second change unifies the agent experience. No earlier than September 28, Copilot Chat on github.com and GitHub Mobile will join the cloud agent under one experience and policy, enabled by default after launch. The cloud agent will continue to run in a Sandbox, while the github.com workflow will move to agent sessions. Users will therefore see fewer independent chat entry points and more of a shared session that can be opened, tracked, and handed off.

Data policy changes alongside the interface. GitHub says the new Copilot agent experience will retain chat data for the life of the account instead of the previous 28-day period. Organizations that opt out will lose the relevant github.com and Mobile features after the new experience launches. For teams with retention, offboarding, regulatory, or customer-code requirements, turning on an agent is therefore also a data-governance decision.

The third change concerns code review. Starting September 28, the Default setting will move from Lite to Balanced effort. Organization or repository overrides, and an explicit user selection of Lite, will remain. That changes the trade-off between latency, cost, and review depth. Administrators should check which repositories depend on Lite for CI timing or usage budgets rather than treating this as only a feature toggle.

The practical meaning is that Copilot administration is moving from separate feature switches toward an agent-use policy. IT teams can document which organizations may enable the cloud agent, which actions need approval, how long chat and code data are retained, who owns overage, and whether the default review effort fits the team's CI window. Those decisions belong in onboarding and offboarding procedures as well as the admin console.

GitHub's post is a future effective-date schedule, not proof that every change was complete on publication day. Rollout may still vary by account type, billing cycle, region, and product stage. Before September 28, the safer operational step is to test agent sessions, Sandbox behavior, data retention, and code review in a test repository, then confirm the result against the organization's policy.

MODULE.002 //

More insights

Ideas on websites, AI automation, digital marketing, AI news, and VMTS updates.