GitHub Suffers Extended Outage Amid a Turbulent August
GitHub has been down for most of the morning of Monday, August 17, 2026, with services degraded since approximately 9:40 AM ET. The platform's status page reported problems with API requests, pull requests, and webhooks, and Copilot was also affected (Engadget).
By 13:40 UTC, GitHub's status page carried the message "Investigating - We are investigating reports of impacted performance for some GitHub services." At 14:04 UTC, the status page updated to report that Pull Requests was experiencing degraded performance with an approximate 20% error rate, and that GitHub was continuing to investigate (GitHub Status). Error rates for web experiences and API traffic hovered around 20 percent, while archive downloads and raw repository content downloads saw approximately 50 percent error rates during the outage (Engadget). As of the Engadget report, the cause had not been determined.
This is at least the second major Microsoft platform outage in recent weeks, following Xbox services being knocked offline for several hours (Engadget). For GitHub specifically, August 17 marks the culmination of a notably turbulent month for the platform's infrastructure.
On August 3, around 22:00 UTC, GitHub's US services were degraded and returned intermittent HTTP 500 errors for API requests, an incident that became the subject of a postmortem (GitHub Status). Two days earlier, on August 1 from 17:47 UTC, users of the Fable 5 model in GitHub Copilot experienced increased request failures and latency (GitHub Status).
The most severe prior incident this month hit GitHub Actions, which is GitHub's built-in automation system for running tasks like testing and deploying code. On August 6, beginning at 15:05 UTC, Actions experienced degraded availability that lasted until 00:14 UTC on August 7. During this window, workflow runs failed or remained queued. At peak severity, 71% of workflow runs experienced infrastructure failures, and 75% of the remaining runs were delayed by more than five minutes (GitHub Status).
The root cause was traced to a routine deployment to an internal Actions service that exposed an existing capacity and concurrency weakness. A latent bug in a GitHub Actions job-assignment service caused runners to be assigned invalid jobs, which they then retried in a loop, preventing them from picking up valid work. Some Actions Runner Controller (ARC) runner pods became stuck in an idle state and remained offline until manually recovered. Additionally, some workflow-triggering events, including push and pull request events, were not processed during the incident and could not be replayed automatically (GitHub Status).
GitHub Actions services recovered at 17:00 UTC on August 6 after the company expanded capacity, throttled incoming webhook-triggered work, and increased processing capacity to clear the backlog of affected events (GitHub Status).
An earlier incident on January 26 saw job failures on some Windows standard hosted runners from approximately 14:03 UTC to 23:42 UTC (GitHub Status). A separate incident degrading data-resident GitHub services was traced to a human error in a configuration change for infrastructure metadata, which modified a specific runtime value (GitHub Status).
The pattern that emerges across these incidents is varied in root cause: a latent bug in job assignment, a configuration error modifying infrastructure metadata, capacity and concurrency weaknesses exposed by routine deployments, and now an undiagnosed outage affecting core platform services including Copilot. The common thread is blast radius. When GitHub Actions degrades, CI/CD pipelines stall and deployments queue indefinitely. When pull requests and webhooks fail, code review and integration workflows break. When the API is down, automation across an organization's toolchain silently fails. And when Copilot is affected, the AI-assisted development layer that a growing share of engineers now depend on for code generation goes dark.
The broader context here is that for teams that have consolidated their SCM, CI/CD, and AI coding assistance on a single platform, each of these incidents is a reminder that platform concentration trades operational simplicity for correlated failure modes. The August 6 Actions incident is particularly instructive: a routine deployment exposed a latent bug and a capacity weakness simultaneously, and the result was that push and pull request events could not be replayed automatically. Teams relying on event-driven workflows had no recovery path without manual intervention.
The cause of the August 17 outage remains under investigation. What is already clear is that GitHub has experienced at least five distinct incidents across its platform in the first two and a half weeks of August, spanning Actions, Copilot, API services, pull requests, and webhooks. For a service that functions as critical infrastructure for a substantial portion of the software industry, that is a density of disruption worth paying attention to.
