# Microsoft 365 and Copilot outage hits Outlook and Teams

> Microsoft says unavailable servers strained its systems, disrupting Outlook, Teams and Copilot worldwide. The BBC reports dependent AI services from OpenAI and Anthropic were also affected.

Published 2026-11-13, updated 2026-11-13. Confidence: medium. When it could matter: Immediate; root cause and recovery stability unknown.

## In plain words
Microsoft had a large outage that hit email, calendars, Teams and its Copilot AI assistant in several countries. Microsoft says the cause was servers becoming unavailable, not an attack. Some AI services from other companies that connect to Microsoft also had problems. If you rely on these tools for work, check whether your team has a backup way to send messages, share files and reach key contacts. Do not assume a restored service will stay up.

## What happened
The BBC reports that Microsoft was working on a global outage reportedly affecting tens of thousands of users. Teams, Outlook and Copilot were affected by what Microsoft called "connectivity problems" [1]. Microsoft posted updates on X, saying it was seeing "positive results" in its restoration efforts.

Downdetector, a monitoring website, recorded reports of Outlook issues peaking globally at nearly 140,000 around 18:00 GMT on Thursday, according to the BBC. The disruption touched Microsoft 365, which businesses use for Outlook, Teams and OneDrive. Microsoft said users might be unable to view certain files or calendar details, and that Copilot might not work properly. Messaging and online meetings were largely functional, the firm added.

Microsoft said on X that the cause was a number of servers "being unexpectedly unavailable", which strained its remaining systems. It later warned that restored services "can become unresponsive again" and that customers "in any region" could be affected. Users in the US, Japan and Singapore reported disruption. OpenAI said it was investigating an issue with its connections to Microsoft's services that made some tools behave unexpectedly. The BBC also says Anthropic was hit. Workhorse Group moved its third-quarter results from 12 November to 16 November, citing business disruption from the outage.

## What is confirmed and what is not
Confirmed by the BBC's account of Microsoft's statements: the outage, the affected products, the stated cause (servers unexpectedly unavailable, straining remaining systems) and the warning that restored services may fail again. Downdetector figures are user reports, not a count of affected accounts.

Claimed or reported only: the BBC says the outage is "reportedly" affecting tens of thousands of users, and that it "understands" there is no indication of a cyber-attack. That is the BBC's understanding, not a published finding by Microsoft.

Unknown from the report: why the servers became unavailable, how long the outage lasted, how many customers lost service, and what OpenAI's and Anthropic's specific problems were. The report contains no detail on Anthropic's response.

## Analysis
This rests on a single report from the BBC, a high-standing outlet, so each claim below stays within what that report supports. The primary documents to read next are Microsoft's own service-status updates and a post-incident review, if one is published.

### Work that depends on Microsoft 365 stops together
The most direct consequence is concentration. Microsoft 365 bundles email, calendars, files, chat and meetings, and the BBC describes it as essential for international organisations. When shared infrastructure fails, one fault can remove several tools at once. Here the sourced effects were calendar and file visibility and Copilot, while messaging and meetings were largely functional. The mechanism Microsoft describes is capacity loss: some servers became unavailable, and the remaining systems were strained. If that is the whole story, the failure is one of load and redundancy, not of data exposure. We are fairly confident of that limit because Microsoft gives no sign of data loss and the BBC reports no indication of an attack, but neither source rules out later findings.

### AI services that sit on top of another vendor
OpenAI said some tools did not work as expected because of problems connecting to Microsoft's services, and the BBC says Anthropic was also hit. This shows that AI products can depend on a cloud or productivity platform run by a different company. Users of ChatGPT or other assistants may see errors even when the AI maker's own systems are healthy. The sources do not say which OpenAI or Anthropic tools failed, or for how long. Our reading is that a dependency map matters: any workflow that chains an AI assistant to Microsoft services inherits Microsoft's failure modes. This is general reasoning, not a finding in the report.

### Repeat failures and unstable recovery
Microsoft warned that restored services could become unresponsive again. The BBC also notes that Microsoft reported an Azure cloud outage the previous month that affected its productivity tools and companies worldwide. The link between the two events is not established in the source; the BBC only places them side by side. A sceptical expert would ask whether they share a cause, and the report does not say. If they did, the pattern would point to a structural weakness, but that is a question for Microsoft's incident reviews. Our earlier coverage of an [earlier Microsoft outage affecting Outlook, Teams and Copilot](/signals/microsoft-outlook-teams-and-copilot-outage-on-12-november) tracks the same event family.

### Business and market effects
The one concrete business effect in the source is Workhorse Group moving its results date, citing business disruptions. That shows a quarterly reporting process disrupted by a vendor fault. The source gives no financial figures and we draw none. For wider context on how such incidents accumulate, see the [incident record since 2018](/record).

### Security: no attack found, but not closed
The BBC's understanding is that there is no indication of a cyber-attack. That is a weaker claim than a confirmed finding. Outages can create openings: staff may switch to unofficial channels, and phishing messages can exploit confusion. That is general knowledge, not something the sources report here. The strongest reason this view could be wrong is that the root cause is still unexplained; evidence of malicious activity in Microsoft's review would change it.

### What would settle it
A Microsoft post-incident report stating the root cause and duration would settle most open points. Statements from OpenAI and Anthropic on what failed and why would clarify the dependency. Downdetector history and Microsoft's service-health history would show whether recovery was stable or repeated. A review linking this event to the Azure outage the BBC mentions would test whether it was one fault or two.

## What can be done
For readers and teams: keep a non-Microsoft route for urgent contact, such as a phone list or a second messaging tool. Save key files and calendar details locally so a short outage does not block work. Treat unexpected messages during an outage with caution and confirm requests through a second channel. See the [Day Zero plan for the day AI services fail](/day-zero) for a wider checklist, and the [live AI risk monitor](/monitor) for current incidents.

For organisations: map which AI tools depend on Microsoft services, set a plan for degraded operation, and ask vendors for incident reviews. Microsoft's own remedy so far, per the BBC, is restoring servers and warning customers that services may fail again.

## Questions readers ask
### What caused the Microsoft outage?
Microsoft said on X that a number of servers were unexpectedly unavailable, straining its remaining systems. The BBC says it understands there is no indication of a cyber-attack. The deeper cause has not been published in the report.

### Which services were affected?
Teams, Outlook and Copilot were affected, as part of Microsoft 365. Microsoft said users might be unable to view certain files or calendar details, while messaging and meetings were largely functional. OpenAI also reported problems with its connections to Microsoft's services.

### How many people were affected?
The BBC says tens of thousands of users were reportedly affected. Downdetector recorded nearly 140,000 Outlook reports at the peak, around 18:00 GMT on Thursday. These are user reports, not a confirmed count of accounts.

### Is it over?
Not necessarily. Microsoft said it was seeing positive results, but warned that restored services could become unresponsive again and that customers in any region could be affected.

## Who is affected

- Businesses using Microsoft 365 for email, calendars and files (relative exposure 9 of 10)
- Users of Copilot in daily work (relative exposure 7 of 10)
- AI providers whose tools connect to Microsoft services, such as OpenAI and Anthropic (relative exposure 6 of 10)
- Companies whose reporting or operations depend on Microsoft tools (relative exposure 5 of 10)
- IT and security teams handling the incident (relative exposure 4 of 10)

## Risks

- Work that depends on Microsoft 365 stalls because several tools fail at once
- AI tools linked to Microsoft services stop working even when the AI maker's own systems are healthy
- Restored services fail again, as Microsoft itself warned
- Confusion during the outage gives phishing and impersonation more room, a general risk the sources do not report
- Root cause stays unpublished, so repeat failures remain possible

## What can be done

- Keep a non-Microsoft channel for urgent contact
- Save key files and calendar details locally
- Map which AI workflows depend on Microsoft services and plan for degraded operation
- Confirm unusual requests through a second channel during outages
- Ask vendors for post-incident reviews

## Sources

1. Global Microsoft outage affecting tens of thousands of users, BBC News, 2026-11-13: https://www.bbc.co.uk/news/articles/c5gm3g9ldz8xo?at_medium=RSS&at_campaign=rss

---
Source: https://www.toswo.com/signals/microsoft-365-and-copilot-outage-hits-outlook-and-teams
Prepared by Toswo Research (https://www.toswo.com/research) from the sources above. Free to cite with attribution: https://www.toswo.com/cite
