Summary
- Anthropic logged a major incident at 09:58 UTC on Friday 13 November 2026 after users reported they could not sign in to Claude.
- claude.ai was listed as a major outage, while Claude Code, the Claude Console and Claude Cowork were also affected. The Claude API stayed operational throughout.
- Anthropic deployed a fix at 11:05 UTC and marked the issue resolved at 11:58 UTC — about two hours after it began.
- The company has not said what caused the failure, how many users were affected, or whether any work was lost.
- The outage is separate from a Microsoft 365 connector problem that has affected Claude since Thursday 12 November.
Claude Went Down on Friday Morning
Claude went down on Friday, and for about two hours many of its users could not get in at all. Anthropic logged a major incident on its status page at 09:58 UTC on 13 November 2026, saying it was investigating an issue with users authenticating to claude.ai and Claude Code. The chatbot itself was not the problem — the models were running — but the front door was shut, and for a service that millions of people now use for work, that was enough to bring the working day to a halt.
The timing made it worse. The outage landed in the middle of the working morning in Europe and the evening in Asia-Pacific: 09:58 UTC was 9:58am in London, 10:58am in Berlin and 8:58pm in Sydney. In New York it was 4:58am, so American users largely woke up to the news rather than living through it. Complaints spiked quickly. Downdetector, which tracks user-submitted outage reports, recorded 2,895 reports about Claude at around 3:45pm Indian time on Friday, against a baseline of just two — a sharp, unmistakable surge.
This was not Claude’s first sign-in failure. Anthropic suffered a similar one-hour outage on 29 September 2026, when it warned that some messages may not have been saved. Friday’s incident was the longer of the two, and the second big sign-in failure in seven weeks — a pattern that will worry the teams and developers who now depend on Claude Code and Claude Cowork for real working days.
What Broke — and What Didn’t
Anthropic titled the incident “Elevated errors on claude.ai, Claude Code, and platform.claude.com”, and its status page listed four services as affected. The first was claude.ai itself, the chatbot on the web and in the mobile apps, which was categorised as a major outage. The second was Claude Code, the company’s coding agent, listed as a partial outage. The third was the Claude Console at platform.claude.com, where developers manage their accounts and API keys, also a partial outage. The fourth was Claude Cowork, Anthropic’s agent for office work, which showed degraded performance.
The fault, as Anthropic described it, was with authentication — signing in, or having an existing login checked. Users reported a range of symptoms, including expired credentials, outright authentication failures, and errors tied to expired OAuth sessions. Some found their sessions had simply stopped working, with no way back in until the incident was over.
What did not break matters just as much. The Claude API itself — api.anthropic.com — was not on the affected list and continued to be listed as operational on Anthropic’s status dashboard. That means the apps and services that other companies build on top of Claude do not appear to have been hit. The failure was in the login layer, not in the models: the inference engine sat behind a sign-in page that many users could not get past.
How Long It Lasted
Anthropic moved relatively quickly once the incident was logged. At 11:05 UTC, about 67 minutes after its first update, the company posted that a fix had been deployed and it was monitoring for recovery. At 11:58 UTC it said the issue affecting claude.ai and Claude Cowork had been resolved — roughly two hours after the incident began.
There was also a second wave to watch. Independent status trackers recorded a further spell of elevated errors later in the day on 13 November, this time including the Claude API alongside claude.ai and Claude Code. Anthropic’s own status page had marked the sign-in incident resolved by then, so the later errors appear to be a separate, shorter disruption — but it underlines that the service had a rough day overall, not a single clean incident.
Was It Linked to the Microsoft 365 Outage?
Some coverage tied Friday’s outage to the Microsoft 365 outage that began on Thursday and has disrupted Outlook, Teams and Copilot for a second day. Anthropic has not made that link, and nothing on its status page supports it. The two incidents ran at the same time, but the company has treated them as separate problems, and there is no evidence that one caused the other.
There is, however, a real Microsoft problem on Anthropic’s status page — just a different one. Since 17:52 UTC on Thursday 12 November, Anthropic has had an open incident for its Microsoft 365 connector, the feature that lets Claude read mail, calendars, SharePoint files and Teams chats. Its latest update says that problem is caused by a Microsoft 365 service incident, that mail, calendar and search tools are still failing for most requests, and that Claude for Outlook remains affected. In other words, even though sign-in is back, the tools that connect Claude to Microsoft 365 may still not work.
The distinction matters for anyone trying to understand what happened. Friday’s login failure was Anthropic’s own incident, with its own cause — still unexplained. The Microsoft 365 connector failure is a knock-on effect of Microsoft’s outage, and it was already running before Claude’s sign-in problems began.
What This Means for Claude Users
For most users, the practical effect was a lost couple of hours: chats they could not open, code sessions they could not start, and a working morning interrupted. Because the API stayed up, the damage was mostly limited to people who use Claude directly, rather than the products built on it. But the incident exposed how much now depends on a single sign-in. A login failure shuts out paying users even when the models themselves are running, and Claude Code and Cowork now carry real working days for many teams.
There are a few practical lessons worth keeping for the next time it happens. First, check the status page before anything else: status.claude.com usually posts within minutes, and if there is an open incident, the problem is not your account, your network or your country. Second, stay signed in. During the September sign-in outage, Anthropic’s own advice was: “If you’re signed in, please don’t sign out.” Signing out to “fix” an error can leave you locked out until the incident ends. Third, check your Claude Code work once you are back in, and make sure changes were saved to your files or repository before carrying on.
The bigger question is the one Anthropic has not answered: what broke, and will it happen again? Two major sign-in failures in seven weeks is a pattern, not a one-off, and until the company explains the cause, users have no way to judge whether the underlying problem has been fixed or merely patched over. For now, the service is back, the login page works, and the working day can resume — but the explanation is still owed.
FAQs
Q1: What happened to Claude on 13 November 2026?
A: Claude suffered a sign-in outage on Friday 13 November 2026. Anthropic logged a major incident at 09:58 UTC, saying it was investigating an issue with users authenticating to claude.ai and Claude Code. Users could not sign in to claude.ai, Claude Code, the Claude Console or Claude Cowork for about two hours.
Q2: How long was Claude down?
A: About two hours. Anthropic posted at 11:05 UTC that a fix had been deployed and it was monitoring for recovery, and at 11:58 UTC it said the issue affecting claude.ai and Claude Cowork had been resolved. Independent status trackers recorded a further spell of elevated errors later in the day.
Q3: Was the Claude API affected by the outage?
A: No. The Claude API (api.anthropic.com) was not on Anthropic’s affected list and stayed listed as operational throughout the main incident. Apps and services built on the Claude API do not appear to have been hit — the failure was in the login layer, not the models.
Q4: Was the Claude outage caused by the Microsoft 365 outage?
A: Anthropic has not linked the two, and nothing on its status page supports a connection. The login failure was Anthropic’s own incident. Separately, a Microsoft 365 connector problem has affected Claude since 17:52 UTC on Thursday 12 November, and Anthropic says that one is caused by the Microsoft 365 service incident.
Q5: What should I do if Claude goes down again?
A: Check Anthropic’s status page at status.claude.com first — it usually posts within minutes. If you are signed in, do not sign out, as Anthropic advised during its September outage, because signing out can leave you locked out until the incident ends. Once back in, check that your Claude Code work was saved to your files or repository.
Related: Claude Fable 5.2 and Mythos 5.2 | Gemini 4 Hydrogen
Sources: Android Authority (13 November 2026); Business Today (13 November 2026); Moneycontrol (13 November 2026); MadRobot (13 November 2026).
