Microsoft Outlook outage enters day two, hitting Claude connectors

More than 130,000 U.S. users reported Outlook problems at the outage's peak on Thursday, November 12, 2026, though Downdetector stressed that the figure represented reports rather than a confirmed count of affected people or businesses.

The Microsoft 365 disruption entered its second day on Friday, November 13, as global recovery continued. Some users still could not access email, while Teams, Copilot and other tools remained unstable. For employers, the immediate people-process issue is straightforward: staff cannot reliably use core communication systems, so internal communications and continuity plans require active testing, not reliance on a recovery announcement.

What changed on Friday morning?

At 6:11 a.m. ET Friday, Microsoft said it had made “additional progress toward recovery” and that a fix was making servers more stable. Some mailbox access and delayed email processing had improved, but intermittent issues remained, and the company gave no timeline for full recovery.

Outlook remained unavailable for some users, while Microsoft's public status page showed service degradation for both business customers and consumer products by Friday morning. The update therefore described partial recovery, not a return to normal operations.

Partial recovery leaves employers unable to assume that every mailbox or workflow is working.

What caused the disruption?

Microsoft first flagged the incident around midday ET Thursday, when users in the Americas struggled to access Exchange Online mailboxes. A follow-up warned that the problem could affect users in any region, extending the scope beyond the initially affected geography.

By Friday morning, Microsoft's public service status page still showed degradation for Microsoft 365 business customers and consumer products. The company said infrastructure faced elevated load across multiple regions because a number of servers had unexpectedly become unavailable, not because customer traffic had increased.

Microsoft restarted servers and reported early Friday that some returned to service could become unresponsive again once traffic reached them. In a 3:34 a.m. Thursday update, it said there was “no indication” the disruption was malicious, a narrower statement than a definitive exclusion of malicious activity.

The stated cause was infrastructure failure, while Microsoft's ability to absorb returning traffic remained unproven.

Which tools and AI connectors were affected?

Microsoft's service advisory said disruption could affect Outlook, Teams, SharePoint, Copilot, Microsoft Defender, Intune, Power BI and other Microsoft 365 products that rely on Exchange Online. That reach matters because one mailbox layer under strain can interrupt several systems used for communications, collaboration and administration.

Anthropic and OpenAI said on their status pages that connectors to Microsoft 365 tools were disrupted. Anthropic also reported authentication issues affecting Claude Code, but did not attribute those issues to the Microsoft outage.

The distinction is material: the source establishes disruption to Anthropic's Microsoft 365 connectors without establishing a shared root cause for Claude Code authentication. Treating the two as one failure would overstate what either company has said.

Claude's Microsoft integrations were affected, while the cause of its separate authentication issue remained unlinked.

How did the outage affect organizations and public services?

Reuters reported that the U.S. House of Representatives told the outlet the outage had affected its users. Electric-truck maker Workhorse Group postponed its quarterly earnings call because of the disruption, showing how a communications failure can delay an executive and investor process even when other operations continue.

Air India said it had rerouted email services, while IndiGo said its email support channel was disrupted on Friday. Both airlines said flight operations were unaffected, an important limit on the scale of the reported disruption.

AP also reported knock-on effects at government agencies in Singapore, the Philippines and Japan. Those disclosures extend the incident across public administration and multiple Asian markets, although the source provides no detailed count of affected agencies or users.

The documented consequences span government, corporate communications and airline support, but not flight operations.

What should employers verify now?

Recovery cannot be judged from a single successful sign-in, especially after Microsoft said restored servers may fail again under load. Employers should test the employee-facing services their continuity plans depend on:

• Can staff send and receive email across regions, rather than merely open selected mailboxes?
• Can Teams and dependent collaboration tools sustain normal use throughout the working day?
• Do employee communications and support channels have working alternatives when primary email fails, a risk illustrated by Air India's rerouting and IndiGo's disruption?
• Can managers access the systems needed for time-sensitive workforce decisions without assuming every Microsoft 365 service recovered together?

Microsoft's explanation also leaves one limit: “no indication” of malicious activity is not the same as an investigation finding. Incident teams should preserve that uncertainty rather than describe the event as either confirmed sabotage or a closed security case.

The next test is sustained stability under returning traffic, not another status label: Microsoft has already reported renewed unresponsiveness, and without a full-recovery timeline, employers should treat Outlook and dependent services as unreliable.