So much of your work now runs on a few cloud providers that a single outage reverberates well beyond the company that caused it. When Microsoft 365 went down on Thursday, it didn’t only take Outlook and Teams offline. It also broke the AI tools you may have connected to them, including Claude’s Microsoft 365 connector and Claude for Outlook. If your AI assistant reads your inbox, checks your calendar or searches your files through Microsoft, it can only work when Microsoft does.
Claude’s Microsoft 365 connector, mail, calendar and search tools is failing for most requests. Claude for Outlook is affected too: If you try to send an email or update your calendar, it might fail to save. and you may not be able to open the add-in at all. Anthropic’s update at 22:23 UTC on November 12, says the problem is caused by a Microsoft 365 service incident and that it will update its post when service is restored. Check the incident page(new window) for the live status.
Is Claude’s Microsoft 365 connector down? What’s happening right now #
Microsoft 365 users began reporting problems around 17:30 UTC (12:30 ET) on Thursday, November 12, and Anthropic opened its connector incident 22 minutes later, at 17:52 UTC. Problem reports on Downdetector(new window) peaked around 130,000, and users in 74 countries reported disruptions. More than 18 hours later, neither company had announced a full fix.
| Time (UTC) | Anthropic’s update on the connector |
|---|---|
| Nov 12, 10:23 PM | Confirmed the cause is a Microsoft 365 service incident. Mail, calendar, and search tools still failing for most requests; Claude for Outlook still affected. |
| Nov 12, 09:18 PM | Microsoft restarting affected servers; failures continue. |
| Nov 12, 08:28 PM | Claude for Outlook also affected: Mail and calendar actions may fail, and some users can’t open the add-in. |
| Nov 12, 06:34 PM | Issue traced to an upstream provider; new connections to the connector may not complete. |
| Nov 12, 05:52 PM | Investigating errors and slow responses for mail, SharePoint, Teams chat, and calendar requests. |
Source: Claude status page(new window)
When the Claude Microsoft 365 connector outage began and how widespread it is #
This incident doesn’t affect Claude itself, only the features that reach into Microsoft.
Microsoft’s own apps took the bigger hit.
| Service | Status |
|---|---|
| Claude Microsoft 365 connector (mail, calendar, search) | Failing for most requests |
| Claude for Outlook | Affected: actions may fail; some users can’t open the add-in |
| Outlook and Exchange Online | Hardest hit: mailboxes wouldn’t open on web, desktop or mobile; emails delayed |
| Copilot Chat , Planner, To Do, SharePoint Online, Power BI, Defender and others | Disrupted, because they depend on Exchange Online |
| Teams | Messaging and meetings mostly working; calendars and scheduling disrupted |
Claude statuses from the Claude status page(new window); Microsoft services as summarized by Proton, citing Business Standard.
What caused the Claude Microsoft 365 connector outage? #
Microsoft did. Anthropic says the connector failures are caused by a Microsoft 365 service incident, so the connector can’t recover until Microsoft does. Microsoft hasn’t published a full root cause, but its updates describe a chain of server failures: A core process became overwhelmed, servers that failed health checks were taken out of service, and the remaining servers then had to carry their load. Microsoft said the extra load came from servers going unavailable, not from more traffic, and that there’s no sign of a security breach or malicious activity.
Is it related to today’s Claude login outage? #
Separately, from 09:58 UTC on November 13, many Claude users couldn’t sign in; Anthropic deployed a fix for that at 11:05 UTC. No link has been confirmed. Anthropic hasn’t explained the cause of the login problem or connected it to Microsoft. The connector incident is the one Microsoft caused, and it was still open as of November 13.
Is your business safe during the Microsoft 365 outage? #
Microsoft says there’s no indication of a breach or malicious activity, so this is an availability problem: You can’t reach your email, calendar or files, and your AI tools can’t either. The bigger risk is scammers. Outages like this tend to bring fake password-reset emails as services come back, so remind your team that Microsoft doesn’t need anyone to reset a password to fix this.
What to do about Microsoft 365 connector outages #
When your AI assistant is plugged into the same provider as your email, files, and calendar, one outage takes all of them down together. Build your setup so that can’t happen.
- Follow live updates on Claude’s incident page(new window) and Microsoft’sstatus.cloud.microsoft(new window) , or in your Microsoft 365 admin center (advisory MO1550821).
- Choose an AI assistant that doesn’t run on the same infrastructure as the rest of your tools. Lumo runs on servers Proton controls in Europe, oninfrastructure independent of Google, Microsoft and AWS . It encrypts saved chats so Proton can’t read them, and doesn’t keep conversation logs or train models on them, which also makes it worth considering when your business handles confidential documents or emails.
- Plan for losing access, not just your data: backups alone aren’t a complete continuity plan . Map which of your tools would fail together, name who in your business makes decisions when something breaks, and set up an out-of-band way to keep email and calls running, such as Proton’sbusiness continuity offering(new window) .
- Watch for phishing. Ignore any password-reset email that arrives as services return, and check directly with your IT team instead.
Spreading your work across tools on independent infrastructure means a single outage can’t stop everything at once.
Keep your business connected through an outage. Over 100,000 organizations worldwide use Proton as their primary workspace. If you’re not ready to ditch Big Tech entirely, you can still get the benefits of Proton for out-of-band communications. Our security experts can help assess your current exposure and make sure outages don’t bring your work to a standstill.