Anthropic's Claude platform went dark for 78 minutes on April 28, taking down Claude.ai, the API, Console, Cowork, and the government product. The outage ran from 17:34 to 18:52 UTC, with authentication errors blocking both human users and programmatic access, specifically impacting the Claude Code tool. By 19:15 UTC, Anthropic confirmed success rates were returning to normal across all services.
The root cause appears to be an authentication system failure. When your identity layer breaks, everything downstream stops working. The API was reachable during the incident but couldn't validate credentials, which is different from a complete infrastructure collapse. Authentication services act as the gateway for every request, and a single point of failure there takes out everything that depends on it. The 78-minute recovery window suggests this required manual intervention rather than automatic failover kicking in.
For companies building on Claude, reliability questions keep piling up. Users on social media and forums have reported multiple Claude Goes Down, Takes Everything With It in recent months despite paying premium enterprise prices. When your government-grade product and your consumer chatbot share the same authentication backbone, one failure hits everyone at once. Anthropic hasn't yet shared a post-mortem explaining what went wrong or what they're doing to prevent a repeat.