# Microsoft 365 outage — Copilot and AI connectors down for 20+ hours

> One Microsoft email failure broke Copilot and the Microsoft 365 tools inside Claude and ChatGPT at the same time.

- Org: Microsoft
- Date: 2026-11-12
- Category: ecosystem
- Source: https://www.theregister.com/saas/2026/11/13/microsofts-massive-m365-outage-stretches-past-20-hours-as-its-fixes-keep-failing/5309609
- Canonical page: https://ai-tldr.dev/releases/microsoft-365-outage-copilot-connectors-nov12/

## In short

A Microsoft 365 outage that began about 18:00 UTC on November 12 took down Exchange Online, Teams and Copilot for more than 20 hours. The Microsoft 365 connectors in Claude and ChatGPT failed with it.

## What it is

Microsoft 365 went down around 18:00 UTC on November 12, 2026, and was still not fully back more than 20 hours later. Exchange Online, Outlook, Teams, Defender, Intune admin dashboards, cloud PCs and Microsoft's own Copilot were hit. Downdetector logged reports from 74 countries within about an hour.

## How it works

The failure sat in Exchange Online, Microsoft's cloud email service: Microsoft said services that interact with Exchange Online were affected, while services that do not need it kept working. Third-party AI tools that read mail, calendars and files through Microsoft's APIs broke as a result. Anthropic first traced its connector errors to an upstream provider, then confirmed it was the Microsoft 365 incident.

## Why it matters

AI agents now act on email, calendars and SharePoint through connectors, so an outage at the data provider stops those agents even when the model itself is healthy. Claude's API and Claude Code stayed operational during the Microsoft incident, yet mail and calendar actions failed for most requests. Teams that run scheduled agent work on Microsoft 365 data need retries and a plan for a day without it.

## Who it's for

teams running AI agents on Microsoft 365 data

## Quick facts

- Started: About 18:00 UTC, November 12, 2026
- Duration: More than 20 hours, recovery still partial
- Microsoft services: Exchange Online, Outlook, Teams, Copilot, Defender, Intune, cloud PCs
- AI connectors hit: Claude and ChatGPT Microsoft 365 connectors, Claude for Outlook
- Stated cause: Servers failing automated health checks; root cause not disclosed
- Cyberattack: Microsoft: no evidence of malicious activity

## FAQ

### Did the Microsoft 365 outage take down Claude and ChatGPT completely?

No. Claude and ChatGPT kept answering chats, but their Microsoft 365 connectors failed. Anthropic's status page said mail, calendar and search requests through the Claude connector were failing for most users and that Claude for Outlook was affected. OpenAI said it was investigating elevated errors affecting its Microsoft 365 connector, where new connections could fail.

### What caused the November 12 Microsoft 365 outage?

Microsoft has not published a root cause. It first described a problem causing servers to fail automated health checks and become unavailable, and TechCrunch quoted the company saying there was no evidence of malicious activity and that the incident was not related to cyber or AI and agents. Microsoft said it was still investigating the cause.

### Was the Claude login outage on November 13 part of the Microsoft incident?

Anthropic did not link them. Its status page lists a separate incident, from 09:58 to 11:58 UTC on November 13, where users could not sign in to claude.ai and Claude Code. That entry names no upstream provider, while the Microsoft 365 connector incident is tracked on its own page and blamed on the Microsoft 365 service incident.

### When did Microsoft 365 and the AI connectors start to recover?

Microsoft reported continued recovery at 13:00 UTC on November 13, with broader improvement in North America and Europe and early signs in Asia Pacific, according to The Register. Anthropic saw signs of recovery for its Microsoft 365 connector and Claude for Outlook at 15:40 UTC, but warned errors could return while Microsoft's incident stayed open.

## Links

- [The Register Report](https://www.theregister.com/saas/2026/11/13/microsofts-massive-m365-outage-stretches-past-20-hours-as-its-fixes-keep-failing/5309609)
- [TechCrunch Report](https://techcrunch.com/2026/11/12/microsofts-outlook-email-hit-by-new-outage)
- [Claude M365 Incident](https://status.claude.com/incidents/sqn3fg1l96dl)
- [The Next Web Report](https://thenextweb.com/news/microsoft-outage-outlook-teams-12-hours-global)
- [Register Analysis](https://www.theregister.com/saas/2026/11/13/microsoft-outage-gives-customers-a-crash-course-in-dependence-on-redmond/5309805)