Image: Anthropic, via Wikimedia Commons (public domain). The Claude logo — the chatbot itself, not a screenshot of the outage.
Key Highlights
- Anthropic logged the first incident at 3:28 PM IST on 13 November 2026: users could not sign in to claude.ai or Claude Code.
- Claude.ai was listed as a major outage; Claude Code and the developer console were partial outages. The Claude API kept working through the first incident.
- Downdetector recorded close to 2,900 user reports at the peak, against a usual baseline of two, with a sharp rise from Indian users.
- A fix was deployed at 4:35 PM IST and the login outage was marked resolved at 5:28 PM IST — about two hours in total.
- A second, separate incident degraded claude.ai, the API and Claude Code from 6:39 PM to 9:08 PM IST. It is also resolved.
What happened?
On Friday afternoon, Claude stopped letting people in. Users reported being signed out in the middle of a session, asked to authenticate again out of nowhere, and then blocked from logging back in. Claude Code users saw 403 authentication errors when the coding tool tried to connect. Anthropic's own status page put it plainly at 9:58 UTC (3:28 PM IST): "We are investigating an issue with users authenticating to Claude.ai and Claude Code."
The fault was in the front door, not the models. The chatbot, the coding tool, the developer console at platform.claude.com and the newer Claude Cowork app all sit behind the same sign-in, so all of them were hit at once. Apps that other companies build on Claude were not affected at first, because the Claude API — the route developers call with an API key — stayed operational through the login outage.
Claude AI down November 13 2026: the full timeline
All times below are Indian time (IST, UTC+5:30) on 13 November 2026, taken from Anthropic's official status page.
| Time (IST) | What happened | Status |
|---|---|---|
| 3:28 PM | Anthropic logs authentication errors on claude.ai and Claude Code; claude.ai is a major outage, Code and Console partial | Investigating |
| 3:20–3:45 PM | Downdetector reports spike to close to 2,900 in India and worldwide, from a usual baseline of two | Reports peak |
| 4:35 PM | A fix is deployed and Anthropic starts monitoring for recovery | Monitoring |
| 5:28 PM | Login outage marked resolved for claude.ai and Claude Cowork — about two hours after it began | Resolved |
| 6:39 PM | Second incident: degraded performance across claude.ai, the Claude API, Claude Code and connectors such as Docs and Sheets | Investigating |
| 9:08 PM | Second incident marked resolved; errors back to normal on all surfaces | Resolved |
| 3:19 AM, 14 Nov | Separate Microsoft 365 connector incident resolved (see below) | Resolved |
Image: Anthropic, via Wikimedia Commons (public domain).
What kept working — and what did not
The two incidents behaved differently, and that distinction matters. During the first, login-only outage, the Claude API was not on the affected list: developers calling Anthropic's models by API key could keep working while anyone who needed to sign in — through the website, the coding tool or the console — hit a wall. The second incident, which began at 6:39 PM IST, was broader: it listed the API alongside claude.ai, Claude Code, the connectors and Cowork, and Anthropic noted that connectors, including Docs and Sheets, stayed degraded even as the main chatbot recovered.
What Anthropic has not said is just as important. The company has not named a cause for either incident, has not said how many people were affected, and has not reported any data loss. The scale we can verify comes from Downdetector, which counts user-submitted reports rather than confirmed affected customers: close to 2,900 reports at the peak, against a normal baseline of two.
This was not Claude's first login failure either. Anthropic dealt with a similar episode of elevated authentication errors on 29 September 2026, also resolved the same day. Sign-in is a single point of failure for a service like this: when authentication breaks, every surface that needs a login breaks with it, even while the models underneath are fine.
Why it matters for India readers
The timing hit India squarely in the working afternoon. The first incident began at 3:28 PM IST on a Friday, when developers using Claude Code and professionals drafting with Claude were mid-task — and Indian users were among the sharpest risers on Downdetector. For anyone who had built a Friday-afternoon workflow around Claude, the outage meant two hours of waiting, not a workaround: there was no alternate login route, and retrying the same sign-in kept failing until Anthropic deployed its fix.
There is also a second thread Indian readers should not confuse with this one. Since Thursday, Claude's Microsoft 365 connector — the add-in that lets Claude read Outlook mail, Teams chats and SharePoint files — had been failing because of Microsoft's own Outlook and Exchange Online outage. Anthropic has said the login failure was a separate issue, and the connector incident was finally marked resolved at 3:19 AM IST on 14 November. If your Claude still cannot read Outlook mail, that is the Microsoft recovery still completing, not the login outage returning.
Should you worry?
FAQs
Is Claude down right now?
No. Both incidents from 13 November 2026 are marked resolved on Anthropic's status page, and the Microsoft 365 connector incident was resolved at 3:19 AM IST on 14 November. If you still cannot sign in, check status.claude.com — a fresh incident would be listed there.
Why couldn't I log in to Claude on 13 November?
Anthropic logged an authentication fault at 3:28 PM IST that stopped users signing in to claude.ai, Claude Code, the developer console and Claude Cowork. A fix was deployed at 4:35 PM IST and the incident was resolved at 5:28 PM IST. Anthropic has not said what caused the fault.
Was the Claude outage caused by the Microsoft outage?
No. Anthropic treats them as separate incidents. The Microsoft 365 outage broke Claude's connector for Outlook mail, calendar and search, while the Claude login failure hit people who never touch Microsoft tools. The two overlapped on the same day, which is why they were easily confused.
Did the outage affect the Claude API?
Not during the first incident: the API stayed operational while sign-in was broken. The second incident later that evening did list the API among the affected surfaces, with degraded performance from 6:39 PM until it was resolved at 9:08 PM IST.
Takeaway
Claude's 13 November outage was a login failure, not a model failure — two hours locked out in the afternoon, a broader wobble in the evening, and no cause named yet. The services are back, the API proved the sturdier route, and the Microsoft connector fault that ran alongside it is resolved too. If you want to understand what the agent features that kept running actually do, read our explainer What Is an AI Agent? Explained in Simple Terms.
Sources
Anthropic status page (status.claude.com) — incident history for 13 November 2026
Business Today — Claude AI is down: Anthropic investigates login failures
Firstpost — Claude AI outage hits users worldwide, login issues reported
JagranJosh — Claude AI Down Today: Anthropic Confirms Login Issues
Gadgets Now — Anthropic Claude AI Down: Users Report Access Issues
MadRobot — Claude down: sign-in outage hits claude.ai and Claude Code
Comments
Post a Comment