Anthropic’s Claude.ai service went dark on April 28, 2026, as users reported outages and authentication failures across its web portal, API, and code playground. The disruption spanned 17:34–18:52 UTC, affecting the main site, Claude Console, and the government‑specific offering. Users across North America and Europe lost access to AI‑driven text generation and automation tools during that window. Resulting in stalled customer workflows and delayed deliverables for several enterprise clients today.

The incident stemmed from an unidentified software bug that inflated request error rates on Anthropic’s API endpoint, api.anthropic.com, and clogged login routes for Claude Code. Engineers traced the issue to a race condition in the authentication module, which triggered repeated 401 responses and forced users to retry repeatedly during the the.

Anthropic’s incident response team rolled out a hotfix within an hour, restoring service stability and reducing error rates below 0.5%. The fix involved patching the authentication logic and tightening rate‑limiting controls. Following the outage, the company pledged to audit its codebase for similar concurrency flaws across all internal components today.

The outage underscores the fragility of high‑scale AI platforms when concurrency bugs slip through testing. Enterprises relying on Claude for code synthesis, policy compliance, and government‑grade security now face tighter reliability demands. Anthropic’s swift resolution demonstrates the importance of real‑time monitoring and rapid rollback capabilities for future deployments today and.