Microsoft 365 is still degraded on Friday morning, 13 November 2026 — a day after it went down worldwide. Microsoft's own status page lists the service as "Service degradation", not restored, with its latest update at 10:01 UTC: recovery is phased, with incremental improvement in several workloads, but service health remains variable and capacity has not returned to baseline. Email, chat, documents and Microsoft's own AI assistant stopped working for a large part of the working world on Thursday afternoon.
The disruption began at around 16:17 UTC on Thursday, first logged as an Exchange Online incident before Microsoft widened it into a broader Microsoft 365 incident later that evening — Microsoft's service-health references are EX1550543 and MO1550821, from its admin-center Service health dashboard (they do not appear on the public status page). What started as an email problem became platform-wide within the hour.
What's down
Microsoft's advisory MO1550821 names five headline services customers may be unable to connect to:
- Exchange Online — Outlook email
- Microsoft Teams
- SharePoint Online
- Microsoft Copilot 365
- Microsoft Defender XDR
That is only the headline list: Microsoft's status page lists 20 affected business services in all — the five above plus the Microsoft 365 admin center, Microsoft 365 apps, the Apps Admin Center, Microsoft 365 Education, Defender for Office 365, Fabric, Graph, Intune, Planner, Power Apps, Purview, To Do, Power BI, Universal Print and Windows 365. Per Microsoft, the common thread is interaction with Exchange Online and email services; features not dependent on Exchange Online remain functional. The consumer side is degraded too: Outlook.com, OneDrive, OneNote, Microsoft To Do, Teams and Microsoft 365 for the web (last updated 10:26 UTC) — for a product family Microsoft says has more than 430 million users.
What Microsoft says happened
Microsoft says a large number of servers became unexpectedly unavailable, leaving the remaining infrastructure overloaded. Servers brought back into service became unresponsive again once they started receiving traffic. Microsoft says its fixes are "not taking as expected" and that it is exploring alternative measures.
No final root cause has been named yet, and Microsoft has given no timeline for full restoration — but it promises a full Post-Incident Report with the root cause once the incident is mitigated. To preserve delayed messages, Microsoft has extended the email timeout window to 48 hours. One thing it has ruled out: the company says it has no indication of malicious activity. This is not a cyberattack, as far as Microsoft can tell.
How big is it
Downdetector, which collects user-submitted outage reports, recorded Outlook reports peaking at roughly 130,000–140,000 globally around 18:00–18:25 GMT on Thursday, spreading to 74 countries and territories within about an hour. Those are reports, not a count of affected users — but they point to a genuinely global event, not a regional blip.
By Friday morning reports had fallen to roughly 28,000–34,000 — better, but still far above normal.
The Azure question
One thing to get right: user reports on Downdetector also named Azure, OneDrive and the Microsoft Store. Microsoft has not confirmed Azure or the Store as affected — its status page explicitly says "No Azure issues detected at this time." OneDrive is a nuance: the consumer OneDrive service is listed as degraded under this Microsoft 365 incident, which is not the same thing as an Azure outage. Azure may or may not have had problems — the evidence available does not establish it, and this article does not claim it was down.
The knock-on: Claude hit twice, in two different ways
The failure spread beyond Microsoft, because so many tools plug into Microsoft 365 — and Anthropic's Claude was hit in two separate incidents that need to be kept apart.
First, the connector incident. From 17:52 UTC on Thursday, Anthropic reported a Microsoft 365 connector incident: mail, calendar and search tools — plus SharePoint and Teams chat — failing for most requests. Claude for Outlook was affected too: mail and calendar actions may fail, and some users cannot open the add-in at all. Anthropic blames the Microsoft 365 outage, and it was still unresolved on Friday morning — bad news for anyone relying on Claude's Microsoft integrations.
Second, a separate login outage. On Friday morning, at around 04:58 ET, Anthropic posted that it was investigating login issues on claude.ai and Claude Code; its status page showed a major outage on claude.ai and partial outages on Claude Code and the developer console. Anthropic has not linked this login outage to Microsoft, and no cause has been stated. Do not treat the two Claude incidents as one.
ChatGPT connectors, too
OpenAI reported an issue with its Microsoft connectors (ChatGPT Connectors/Apps): new connections may fail and existing tools may not work as expected. That issue had been ongoing for roughly 11 hours as of Friday morning. Like Claude's connector problem, it is downstream of the same outage.
What is not affected
Not every cloud was falling. Amazon said its services were operating normally and that reports linking AWS to the disruption were inaccurate. Unconfirmed user reports also named Google, Cloudflare and Steam; there is no confirmation for any of those.
Real-world impact
Beyond office inboxes: Singapore's Government Technology Agency said government emails were down. The Philippines' Supreme Court said the outage affected some of its operations, including email responses. Workhorse Group delayed a quarterly earnings call. Australian reporting named the Australian Federal Police, ASIO and the ABC among those affected.
What to watch
Three things decide how this ends. First, whether the phased recovery holds — health is still variable, capacity is not back to baseline, and Microsoft's alternative measures have to keep servers in service once traffic returns. Second, whether a root cause is named: until it is, nobody can judge whether this can happen again tomorrow. And third, whether the downstream incidents at Anthropic and OpenAI clear once Microsoft 365 recovers, or need their own fixes. Check Microsoft's and Claude's status pages — and until Microsoft says the incident is resolved, treat every "back to normal" report with caution.
Sources
Checked 13 November 2026:
- Microsoft 365 status page, read first-hand 10:35 UTC 13 November 2026 — https://status.cloud.microsoft
- Anthropic status page — https://status.claude.com
- OpenAI status page — https://status.openai.com
- Associated Press reporting on the Microsoft 365 outage
- Reuters reporting on the Microsoft 365 outage
- Downdetector user-submitted report figures
- The Wall Street Journal reporting on the outage
- BBC account of Downdetector data