GitHub experienced a major global outage beginning around 13:40 UTC on August 17, 2026, that degraded Pull Requests, Issues, Actions, Webhooks, and Copilot for several hours. Error rates climbed to roughly 20% across general web traffic and the API, while archive downloads and raw file access, endpoints many CI pipelines and package tooling depend on, saw error rates near 50%. Downdetector logged more than 3,000 user reports at the peak, with the heaviest concentrations in the US and India. GitHub restored seven of the eight affected services (API, Actions, Git operations, Issues, Pages, Pull Requests, and Webhooks) to operational status by 16:59 UTC, but Copilot remained in a major-outage state past that point, and the company had not yet disclosed a root cause. For any team whose CI/CD, code review, or deployment pipeline routes through GitHub, an outage of this duration is a direct productivity hit: merges stall, automated tests queue up, and release pipelines that gate on GitHub Actions simply stop. It is also a reminder that centralizing so much developer workflow, hosting, review, automation, and now AI pairing, on a single vendor creates a single point of failure whose blast radius keeps growing. The outage's timing was notable: it landed the same day a competitor unveiled its own code-hosting platform, giving that launch unusually sharp real-world contrast. Teams with hard release deadlines should have a documented fallback for CI/CD interruptions, such as mirrored repos, alternate CI runners, or manual deployment paths, rather than assuming platform uptime.