GitHub Outage Spikes Error Rates, Cripples SSO and Lingers in Copilot Auth

GitHub suffered a broad platform outage that degraded most of its core services at once, including Git operations, the REST/GraphQL API, Actions, Pages, Issues, Pull Requests, Webhooks, and Copilot. At peak, web and API traffic saw roughly a 20% error rate, while archive and raw repository content downloads failed about half the time. Critically for enterprise customers, single sign-on took a direct hit: SAML and OIDC authentication, SCIM provisioning, and Team Sync were all impaired, meaning some organizations couldn’t authenticate or manage access during the window.

Engineers traced the disruption to a single faulty component and applied corrective action, but recovery was messy. Even after the main failure was mitigated, sporadic authentication errors persisted across services. GitHub partially disabled authentication token retries as a mitigation and monitored the effect before rolling it out fully. The longest-lingering symptom was Copilot authentication failing intermittently in some third-party applications, though Copilot access through the GitHub CLI and GitHub App kept working throughout.

The incident was ultimately resolved, with GitHub promising a detailed root cause analysis once available. The episode is a reminder of how tightly modern development pipelines depend on GitHub as a single point of failure — when its auth and API layers wobble, everything from CI/CD runs to enterprise SSO logins and AI coding assistants degrades in tandem.