Anthropic investigates Claude outage affecting thousands of users worldwide
In Focus
- Claude went down for thousands of users on June 23, 2026, starting around 10:02 AM ET
- Anthropic confirmed the issue, saying it was actively investigating the elevated error rate
- The outage affected all Claude models and platforms, except Claude for Government users
- Reports on Downdetector peaked above 8,000 in the US before gradually declining.
Anthropic’s AI assistant Claude experienced a significant service disruption on June 23, 2026, leaving thousands of users unable to access the platform. The company acknowledged the issue through its official status page, confirming an elevated error rate across multiple models while working to restore full service for affected users.
What happened during the outage
The disruption began at approximately 10:02 AM ET, when a sharp spike in user complaints surfaced on Downdetector, with reports in the United States exceeding 8,000 at their peak. The Claude outage was not limited to a single model or interface. Users reported that Claude was either stuck in an indefinite thinking loop or returning a message stating the model was unavailable.
Both the web version and the Claude Code interface were affected, impacting free and paid subscribers alike. Anthropic status page data confirmed a major service disruption across its chat interface, Claude Code, Claude Cowork, and its API, with the only exception being Claude for Government. As downdetector reports climbed, Anthropic posted updates indicating that the issue had been identified and that a fix was being implemented, though service remained unstable for more than an hour after the initial reports surfaced.
A broader and more severe disruption
This Claude outage was notably larger than a previous incident just three weeks earlier on June 2, which had been isolated to the Claude Opus 4.6 model. By contrast, the June 23 event represented a wider ai service disruption spanning all available models and platforms simultaneously. Anthropic’s own uptime history, as reflected on its status dashboard, showed a pattern of recurring yellow and red indicators, pointing to ongoing reliability challenges for the platform.
Engineers worked to push a fix, and while downdetector reports began to fall from their peak, many users continued to experience problems well into the afternoon. The company stated it was continuing to monitor results after implementing its patch, though the platform had not fully stabilized at the time initial reports were filed. The incident renewed questions about the resilience of AI infrastructure as reliance on tools like Claude grows across both individual and enterprise use cases.
This Claude outage also highlighted how dependent developers and professionals have become on uninterrupted access to Claude models down scenarios presenting real workflow consequences. Anthropic has not yet published a full post-incident report.
