GitHub GitHub Actions CI/CD DevOps outage

GitHub Actions Went Down for Three Hours and Told Developers Their Accounts Were Suspended

29 May 2026

GitHub Actions experienced an outage lasting over three hours on May 26, disrupting CI/CD workflows for developers worldwide and displaying a misleading "Your account is suspended" error message, which caused additional alarm given how difficult real account suspensions can be to resolve. The outage, attributed to authentication issues, was particularly disruptive because even customers using external or self-hosted runners were affected, as GitHub's cloud service acts as the control plane. Despite recurring reliability problems this year, GitHub's platform continues to grow rapidly, with Actions usage more than doubling since 2023, largely driven by the surge in AI-generated code.

GitHub Actions fell over for more than three hours on May 26th, blocking CI/CD pipelines and greeting developers with a deeply unhelpful error message claiming their accounts had been suspended. They hadn't been. It was just a bog-standard authentication failure dressed up in the most alarming language possible.

The first user complaints surfaced around 1030 UTC. GitHub's official incident report kicked off at 1057 UTC with the characteristically understated description of "degraded performance for Actions and Pages," before being quietly upgraded to acknowledge that the majority of Actions runs were affected.

For anyone whose job involves keeping a CI pipeline green, three hours of this is genuinely painful. Unlike a repository access outage, where you can at least keep working locally, a broken Actions service has no easy workaround. If your build and deployment workflow runs through GitHub, you're stuck. "Our CI is currently basically blocked," wrote one developer on call for their company's continuous integration team. Not great.

The suspended account message made things worse. Actual account suspensions by cloud providers are a bureaucratic nightmare, sometimes taking months to untangle. One developer in the resulting discussion thread mentioned having their GitHub account suspended for four months before support eventually admitted it was a mistake. So when your failing build throws up "Sorry. Your account was suspended," the heart sinks before the brain catches up.

Self-hosted or external runners offered no escape either. GitHub's cloud infrastructure acts as the control plane regardless of where the runner VMs live, so even teams who'd done the work to host their own execution environments were caught out.

GitHub marked the incident resolved at 1318 UTC, though it noted that a small number of issues, pull requests, comments, and discussions had been incorrectly marked as hidden, with remediation still ongoing.

This isn't an isolated wobble. GitHub reliability has been patchy throughout 2025 and into 2026. Some of it can be pinned on AI agents and LLM data scrapers hammering the platform. Authentication failures are harder to explain away that easily.

After every outage the predictable conversation kicks off about alternatives, self-hosted Git, Forgejo, GitLab, roll your own. And every time, most organisations stay put. The free tier is generous, migration is expensive, and existing workflows don't move cheaply.

The platform is also growing fast, which probably explains some of the strain. GitHub's COO Kyle Daigle noted last month that the platform hit one billion commits in all of 2025, and is now clocking 275 million commits per week, putting it on pace for 14 billion this year. GitHub Actions has gone from 500 million minutes per week in 2023 to over 2.1 billion minutes in a single week recently. AI-generated code is almost certainly responsible for a chunk of that growth, producing output at a volume no human team could match.

More traffic, more failures. The maths isn't complicated.