Anthropic's Claude platform went down for roughly two hours on Friday morning, blocking users from logging in to claude.ai and the company's developer tools before engineers restored service. According to Anthropic's official status page, the incident began shortly before 10:00 UTC and was fully resolved by 11:58 UTC on November 13.

The company's incident report, titled "Elevated errors on claude.ai, Claude Code, and platform.claude.com," listed four affected products: claude.ai, the Claude Console (platform.claude.com), Claude Code, and Claude Cowork. The disruption centered on authentication — users were unable to sign in — rather than on the behavior of the models themselves. For more context on this story, see our ongoing AI trends.

What happened, minute by minute

The status page timeline was short but specific:

  • 09:58 UTC — Anthropic opened the incident: "We are investigating an issue with users authenticating to Claude.ai and Claude Code. We will provide an update as soon as possible."
  • 11:05 UTC — the company reported that "a fix has been deployed for the issue affecting Claude services and we are monitoring for recovery," moving the incident to a monitoring state.
  • 11:58 UTC — the incident was marked resolved, with the status page stating that "the issue affecting claude.ai and Claude Cowork has been resolved."

That is a total outage window of exactly two hours from first report to resolution, with a fix deployed inside the first hour. The status page did not publish a root-cause analysis as of the time of writing, and the incident report did not indicate any data loss; it described a service availability and authentication failure.

An authentication problem, not a model failure

The language Anthropic used matters. The incident was framed around users "authenticating" to the services — the login and session layer that sits in front of Claude's products. During the window, requests surfaced elevated errors across the consumer app, the developer console, and the API-adjacent platform domain listed in the incident title.

For a company whose revenue increasingly depends on API access and developer tooling, an authentication-layer failure is in some ways the most disruptive kind: it blocks everyone, regardless of which model or feature they are trying to reach, and it does so at the front door.

Why Claude Code outages sting developers hardest

Claude Code, Anthropic's terminal-based coding agent, has become a daily driver for a large share of professional developers, and it was explicitly named in the incident report. When login systems fail, work in progress stalls in a particular way: long-running refactoring sessions pause mid-task, and any workflow that needs to re-authenticate to the platform simply cannot proceed.

The developer community's reaction was immediate. On Hacker News, a thread titled "Elevated errors on claude.ai, Claude Code, and platform.claude.com" linking to the status page accumulated 266 points within hours, while a parallel "Claude Is Down" thread reached 129 points. Whatever the eventual post-mortem says, the volume of that discussion is a rough proxy for how many engineering teams felt the outage in real time.

A bruising stretch for cloud reliability

The Claude incident landed less than a day after a far larger failure elsewhere in the stack. Microsoft spent Thursday fighting a global outage of Microsoft 365 services including Outlook and Teams, with CNN running live coverage of the disruption and the BBC reporting that the company said attempts to fix it were not going "as expected." A Hacker News thread tracking the outage passed 360 points while users reported Microsoft 365 still down after twelve hours.

Two outages at two different vendors in two days do not share a cause, and nothing in the public reporting suggests they do. But together they underline the same structural fact: knowledge work — and now AI-assisted knowledge work in particular — concentrates on a short list of cloud services whose failure is felt instantly across millions of desks.

The reliability bar rises as AI moves into production

The episode also illustrates how the economics of downtime change when AI agents enter the workflow. A chatbot being down is an annoyance; a coding agent that many teams keep running through their working day is closer to build infrastructure, where minutes of unavailability translate directly into interrupted tasks and idle engineers.

For enterprises, the practical responses are familiar from the broader cloud era: service-level agreements that acknowledge AI tooling as critical infrastructure, monitoring that watches vendor status pages rather than just internal dashboards, and — increasingly common — fallback routing that can switch work between competing models when a primary provider degrades. Anthropic's own transparency here is consistent with industry practice: a public, granular timeline per product, with email and SMS subscription options for updates.

What to watch

Anthropic's incident report stands at three states — investigating, monitoring, resolved — with no published root cause so far. Whether the company follows up with a deeper post-incident analysis, as it has done after previous disruptions, will be worth watching for teams that build on Claude products.

For now, the practical takeaway is simple: service was restored within two hours, all four affected products are back according to the status page, and no impact on model behavior or data was reported. The outage is a reminder, arriving on the heels of Microsoft's own troubles, that even the infrastructure behind frontier AI is still just infrastructure — and it fails like infrastructure does.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →