Top News

Anthropic’s AI platform Claude experienced a prolonged partial outage on March 2, lasting approximately 2 hours and 45 minutes, according to the company’s status page.

The disruption, categorised as a “Partial outage,” was linked to elevated errors across multiple services, including claude.ai, the console, and Claude Code. The status page also showed elevated errors impacting specific models such as Claude Opus 4.6 and Claude Haiku 4.5, indicating that the issue extended beyond just the web interface.

This marks a significant update from earlier reports that suggested a shorter disruption window. The extended downtime meant users across both web access and certain model endpoints were affected for a considerable period.

The following day, on March 3, the platform again reported “Elevated errors in claude.ai,” with the issue marked as “Investigating.” The recurrence of issues within 24 hours has raised concerns among users who rely heavily on Claude for development, writing, research and enterprise workflows.

While Anthropic clarified during the March 2 incident that the Claude API was functioning as intended at one stage of the investigation, the broader status logs show that multiple components and models experienced elevated errors during the outage window.

Back-to-back disruptions of this scale are likely to intensify scrutiny around uptime reliability, infrastructure scaling and service resilience as generative AI platforms become increasingly embedded in business and professional environments.

Anthropic has not yet released a detailed post-incident technical explanation for either disruption.

Further updates are awaited as the March 3 investigation continues.