A Claude outage hit users twice across less than a day, disrupting access to Anthropic’s consumer-facing AI services and triggering a wave of login failures and error screens. Reports described a first disruption on March 2 that lasted for several hours in multiple regions, followed by a second interruption on March 3 with fresh availability problems.
Anthropic’s public status updates said the issue was under investigation from 03:15 UTC on March 3, with follow-up notes continuing into the 04:43 UTC update window. Users said they could not sign in and repeatedly saw messages such as “Claude will return soon” or prompts to try again later.
Key details reported during the incident:
- Impacted surfaces included claude.ai, associated apps, the developer console and the Claude Code tool
- Outage trackers cited thousands of U.S. reports during the March 3 disruption, plus hundreds in India
- Coverage diverged on operational scope, with some accounts suggesting API-linked impact while another explicitly said the API continued functioning
