Repeated GitHub outages raise reliability concerns for its AI coding tools
- GitHub’s status records and monthly availability reports show repeated outages in February, March, April and May 2026 that hit core services, including Copilot. - GitHub said more than 180 million developers now use the platform, while a March incident drove github.com request failures to about 40%. - GitHub posts incident updates on GitHub Status and monthly postmortems on its blog, including April 2026 availability details.
GitHub’s own status records show a run of outages in 2026 that hit not only repository and API access, but also the company’s AI coding products. February, March, April and May incidents affected combinations of github.com, Git operations, Actions, Codespaces and GitHub Copilot, according to GitHub Status and the company’s monthly availability reports. GitHub has said more than 180 million developers now use the platform, a scale that raises the stakes when interruptions affect coding, reviews and automated workflows. In AI coding tools, where suggestions are expected to appear inside active development sessions, those interruptions have become part of the competitive discussion around Microsoft-owned GitHub. ### Which outages are fueling the reliability debate? GitHub’s February 2026 availability report said two incidents on February 9 affected github.com, the GitHub API, GitHub Actions, Git operations, GitHub Copilot and other services. The company said the first disruption ran from 16:12 UTC to 17:39 UTC and the second from 18:53 UTC to 20:09 UTC. GitHub’s March 2026 availability report said a March 3 incident again hit github.com, the API, Actions, Git operations, Copilot and dependent services. At the peak, GitHub said github.com request failures reached about 40%, while roughly 43% of API requests failed. GitHub’s April 2026 availability report said users were unable to use Copilot Chat on github.com and Copilot Cloud Agent between 15:16 UTC and 19:18 UTC on April 22. (github.blog) The company said Copilot Memory, then in preview, was also unavailable to Copilot agent sessions during that period. ### How recent were the latest disruptions? GitHub Status showed an Actions degradation on May 20, 2026 from 16:58 UTC to 20:14 UTC. (github.blog) The status history also recorded a May 15 Actions incident caused by a planned failover of supporting infrastructure, which GitHub said led some workflow runs to fail or start late. GitHub Enterprise Cloud’s U.S. status page showed a separate May 6 Copilot-related incident in which users could not start or view Copilot Cloud Agent or remote sessions between 11:02 UTC and 11:13 UTC. (github.blog) GitHub said requests to the session API returned errors, preventing users from creating new sessions or viewing existing ones. ### Why do these outages matter more for AI coding than for ordinary downtime? (githubstatus.com) GitHub said in its latest Octoverse report that “180 million-plus developers” now work and build on the platform. The company also said the release of GitHub Copilot Free in late 2024 coincided with a jump in sign-ups, adding to usage across repositories, pull requests and code pushes. (us.githubstatus.com) GitHub’s own Octoverse material said software teams are increasingly focused on “reducing friction” as AI becomes part of development workflows. That makes availability and response times more central to products such as Copilot Chat and Cloud Agent, because those tools are used during active coding sessions rather than in separate, occasional tasks. That connection between workflow friction and tool performance is an inference drawn from GitHub’s product positioning and the incident record. (github.blog) ### Has GitHub said what it is doing about reliability? GitHub’s February availability report said the company added new monitors and alerts after the February 9 incidents to reduce detection time and prevent similar failures. The January availability report said GitHub was also working to improve validation for upgrades that fail only under high load, while reducing mitigation times. (github.blog) Vladimir Fedorov, GitHub’s chief technology officer, said in an April blog post titled “An update on GitHub availability” that reliability work was a company priority. The post identified availability as part of GitHub’s effort to support developer productivity as usage grows. ### Where can developers track the next incident or postmortem? GitHub runs a public status site with real-time incident updates and a history page listing resolved disruptions across services including Actions and Copilot. (github.blog) The company also publishes monthly availability reports on its blog with dates, durations, affected products and remediation steps. GitHub’s next formal accounting of May incidents is likely to appear in its May 2026 availability report on the company blog, following the pattern of January through April reports already posted. (github.blog 1) (github.blog 2) (githubstatus.com)