Anthropic’s Claude AI platform experienced a partial outage on September 29, 2026, affecting multiple services across the platform.
According to Anthropic’s status page, the incident began around 14:21 UTC when elevated error rates were reported affecting Claude.ai (including desktop and mobile apps), Claude Code, and Claude Cowork. Users reported failed requests, errors loading or sending conversations, and being asked to sign in again.
The outage expanded shortly after. By 14:28 UTC, the Claude API was also returning errors for some requests. At 14:30 UTC, Anthropic reported investigating degraded performance across claude.ai, platform.claude.com, the Claude API, Claude Code, and Claude Cowork.
A first mitigation was applied at 14:36 UTC, reducing error rates substantially, though some sign-in attempts, account actions, and Claude Code and Cowork sessions continued to fail. At 14:41 UTC, error rates had dropped substantially but issues persisted.
Most services recovered by 14:59 UTC. According to the status page, signing in, starting new chats, voice conversations, Claude Code and Cowork sessions, purchases, and file uploads were working again. However, Anthropic noted that some messages sent between 14:00 and 14:59 UTC may not have been saved.
Anthropicadvised users who were signed in to remain signed in, as a second issue had prevented many users from signing in via single sign-on and Sign in with Apple.
The company indicated it was continuing to monitor the situation closely and would provide further updates within 30-minute intervals. By 15:30 UTC, all services had been operating normally for approximately 30 minutes, with Anthropic continuing to monitor for any recurrence.
Key facts
- Outage began around 14:21 UTC on September 29, 2026
- Affected Claude.ai, Claude Code, Claude Cowork, the Claude API, and platform.claude.com
- Most services recovered by 14:59 UTC, within approximately 38 minutes of the initial reports
- Some messages sent during the outage window may not have been saved
- Sign-in issues via single sign-on and Sign in with Apple persisted into the recovery phase