Claude login outage: your AI control room has only one door

The office was lit, the desks were tidy, everything inside was working. The door just wouldn't open. That's roughly what Friday morning in Europe felt like for anyone who runs their work through Claude: projects, briefs, skills, agents in Claude Code and Cowork. And almost nobody has a spare key.

What happened

According to Anthropic's status page, the team started investigating an issue with users authenticating to claude.ai and Claude Code at 09:58 UTC. Listed as affected: claude.ai, the Console (platform.claude.com), Claude Code and Claude Cowork. A fix was deployed at 11:05 UTC, and the incident was marked resolved at 11:58 UTC. The Claude API isn't mentioned in the notice; in the November 8 incident, by contrast, Anthropic named it explicitly.

The timing matters. Two hours in the middle of the European workday, while it was 2 to 4 a.m. in California. It's also the second Claude disruption in under 24 hours that teams here actually felt: on Thursday evening, the Microsoft 365 outage took down Claude's Outlook and calendar connections. Yesterday it was the inbox. Today it was the door.

The control room is the asset, not the model

I've been saying it since late October: your own Claude project is your control room. Strategy, briefs, drafts, one project with memory per campaign. I still believe that. But a control room with a single door is a risk almost no team has on its radar.

Sounds like an IT problem. It isn't. What lives in those projects is working knowledge: tone of voice, positioning, no-gos, the prompts that actually work. That's an owned asset sitting in someone else's building. When the login fails, the AI isn't gone. Access to your own know-how is.

Here's what I find telling: the pattern isn't new. During the March 2 outage, reports said many users were stuck at login while Anthropic said the API was working normally. So in my view, teams that run their critical workflows through an API key had a decent chance of not noticing today at all. Teams that do everything in the chat window got a two-hour break.

The catch

A spare key takes work. API access needs someone to set it up, watch costs and manage credentials properly. AI doesn't fix broken ways of working; it widens the cracks. If your know-how has never been written down outside the chat, days like today are when you find out.

What I'd do now

• Export project instructions, briefs and your best prompts as files this week, into a folder you own. The knowledge belongs to you, not to a login.
• Flag the two or three workflows that can't wait two hours, and give exactly those an API route or a second tool.
• Name one person who decides during an outage: wait, switch or go manual. That call stays in-house.

Bottom lineYour know-how belongs in your folder, not behind a login you don't control.