GitHub experienced a series of simultaneous service failures on February 9, 2026, affecting Actions, pull requests, notifications, and its AI assistant Copilot, according to a report by Richard Speed published in The Register on February 10, 2026. The Microsoft-owned platform acknowledged a multi-service disruption at 1554 UTC and did not confirm full restoration until 1929 UTC — a window of nearly four hours.
- On February 9, 2026, GitHub’s Actions, pull requests, notifications, and Copilot all suffered outages simultaneously.
- Notification delays reached approximately 50 minutes; GitHub declared operations restored at 1929 UTC, roughly four hours after the incident began.
- Copilot policy propagation failures persisted from 1629 UTC on February 9 to 0957 UTC on February 10 — a span of more than 15 hours.
- Unofficial reconstructions of GitHub’s status feed show uptime dropped below 90% at one point in 2025, well short of the 99.9% figure in its Enterprise Cloud SLA.
What Happened
On February 9, 2026, GitHub’s Actions, pull requests, notifications, and Copilot services all experienced disruptions. The platform publicly acknowledged problems with “some GitHub services” at 1554 UTC. By 1757 UTC, notification delays had narrowed from roughly 50 minutes to approximately 30 minutes, and GitHub declared full restoration at 1929 UTC.
A separate Copilot-specific failure ran considerably longer. GitHub reported problems with Copilot policy propagation from 1629 UTC on February 9 through 0957 UTC on February 10 — a period of more than 15 hours. As GitHub described it: “This may prevent newly enabled models from appearing when users try to access them.”
Why It Matters
GitHub’s Enterprise Cloud Service Level Agreement commits the company to 99.9% uptime — commonly called “three nines.” However, as Richard Speed noted in The Register, this SLA does not extend to all users on the platform, meaning many developers who experienced the February disruption had no contractual recourse.
The February incidents did not occur in isolation. Unofficial reconstructions of GitHub’s public status feed — which Speed noted “requires caution” as an unofficial source — indicate that GitHub’s uptime dropped below 90% at one point during 2025. The platform also changed its official status page format prior to February 2026, a move Speed described as making it “harder to visualize the availability of its services” over 90-day periods, leaving customers with limited visibility into cumulative reliability data.
Technical Details
The February 9 multi-service incident produced notification delays of approximately 50 minutes at peak. GitHub’s own incident timeline showed delays improving to approximately 30 minutes by 1757 UTC, before the company declared full restoration at 1929 UTC — roughly three hours and 35 minutes after first acknowledging the disruption.
The Copilot disruption involved policy propagation failures affecting “some users,” preventing newly enabled models from loading. The failure window stretched from 1629 UTC on February 9 to 0957 UTC on February 10, covering overnight hours across most major development hubs in Europe and the Americas.
On the broader reliability picture, Speed noted that five nines — 99.999% uptime — represents the industry gold standard, equating to approximately five minutes and 15 seconds of allowable downtime per year. GitHub’s Enterprise SLA targets 99.9%, permitting roughly eight hours and 46 minutes of annual downtime. A dip below 90% uptime, as recorded via the unofficial status feed for 2025, would permit more than 36 days of outage in a calendar year.
Who’s Affected
Enterprise Cloud customers with active SLA agreements were the most directly affected, as they hold the contractual expectation of 99.9% uptime. Developers using GitHub Actions for continuous integration and deployment pipelines faced workflow interruptions during the roughly four-hour window on February 9. Those relying on Copilot — particularly teams that had recently enabled new models — were unable to access those features for more than 15 hours.
The disruptions carried downstream consequences for any software teams using GitHub as their primary development platform, given how tightly Actions and Copilot are integrated into modern build and review workflows. GitHub’s SLA coverage gap means the majority of its user base had no formal remediation path.
What’s Next
As of Speed’s February 10 report, GitHub had not announced specific measures to address the reliability issues. The company’s decision to restructure its status page has reduced the visibility customers have into cumulative 90-day uptime, leaving unofficial feed reconstructions as the primary tool for assessing historical availability.
The February outages arrived amid several other pressure points for the platform: GitHub had separately been weighing a kill switch on pull requests to limit AI-generated code submissions, and a security incident involving the Shai-Hulud worm had exposed secrets across roughly 25,000 repositories, according to related reporting linked in The Register’s article. Author details beyond Richard Speed were not available at time of publication.
