The Claude spark symbol. Image: Anthropic
Claude suffered a major outage on Tuesday, knocking out the chatbot, Claude Code, Cowork and the Claude API for about an hour. Anthropic says on its status page that “some messages sent between 14:00 and 14:59 UTC may not have been saved”.
What went wrong #
The trouble started at 14:00 UTC (10am in New York, 3pm in London). Anthropic first reported “elevated error rates” across Claude.ai, including the desktop and mobile apps, Claude Code and Claude Cowork, with some API requests also failing. Users saw failed requests, conversations that wouldn’t load or send, or were asked to sign in again. Reports on Downdetector spiked at the same time.
Anthropic applied a fix at 14:36 UTC and error rates dropped, but a second problem followed. According to the status page, it stopped many people from:
- Signing in , with single sign-on and Sign in with Apple unavailable;
- Starting anything new : new chats, voice conversations, and Claude Code or Cowork sessions;
- Paying or up : purchases and file uploads failed.
At one point Anthropic’s advice was simple: “If you’re signed in, please don’t sign out.”
Is Claude working now? #
Yes. Anthropic says most services recovered at 14:59 UTC, and at 15:30 UTC it said everything “has been operating normally since 14:59 UTC”, though it is still monitoring. The outage hit claude.ai, the Claude Console, the Claude API, Claude Code and Cowork, so developers’ apps built on the API were affected as well as people using the chatbot.
If you got a “sign in” prompt or an error during that hour, it was the outage, not your account or your country. (If Claude says it isn’t available where you are, that’s a different problem: our guide to Claude’s supported countries explains it.)
What to do if a chat is missing #
Because messages sent between 14:00 and 14:59 UTC may not have been saved, check any conversation you were in during that hour. If a reply or your own message is missing, send it again. For Claude Code, it’s worth checking that any work done in that window was actually saved to your files or repository before carrying on.
Anthropic hasn’t said what caused the outage. It came a day after the launch of Claude Sonnet 5.5, but the company hasn’t linked the two.
Why it matters #
Millions of people and businesses now run their work through Claude, and many developers build their own products on its API, so an hour offline ripples a long way. Lost messages make it worse than a simple slowdown. Anthropic will be under pressure to explain what broke.