Microsoft 365 outage disrupts Outlook and AI services globally

The scope of the disruption

On Thursday, November 12, 2026, a significant outage struck Microsoft 365, the company's primary subscription bundle for workplace applications. Users across the globe reported being unable to open Outlook inboxes, while other features in Microsoft Teams were delayed or disrupted. The incident was widespread enough that it affected multiple regions simultaneously, including the Americas and Australia.

Data from Downdetector showed that reports of Outlook problems peaked at 138,058 in the United States around 1 p.m. Eastern time. While the numbers began to fall by the evening, many users still faced issues lasting more than five hours. The disruption was not limited to a single country, as evidenced by reports from Australia where Service NSW had to ask customers to delay visits because transactions were unavailable.

The outage was specifically linked to Exchange Online, which is the backend service responsible for storing and delivering work emails and managing calendars. Because Outlook is the primary app people use to access this data, the app might still open on a user's device even if the inbox fails to load. This separation between the user interface and the underlying data service explains why some users could still interact with parts of their software while others were completely locked out.

How the servers failed

Microsoft identified the root cause as a failure in the software that establishes secure connections on the affected servers. According to the company, this software became overwhelmed, causing servers to drop out of the network. As these machines failed, the remaining servers were forced to handle a higher volume of incoming requests, leading to a cycle of lost capacity and further crashes.

The company is currently investigating whether a specific change to its network settings contributed to the initial failure. To address the issue, engineers began reversing that recent change and started testing a fix on a small group of machines. The goal of this process is to ensure that repaired servers can continue to respond even as normal traffic levels return to the system.

A spokesperson for Microsoft told The Guardian that the company has seen no indication of a cyberattack during this event. This distinction is important because it suggests the issue was a technical malfunction rather than a malicious breach. The company is still working to establish exactly what triggered the software to become overwhelmed in the first place.

Impact on AI and Teams

The outage had a notable ripple effect on artificial intelligence services that integrate with Microsoft 365. Tools like ChatGPT and Claude use connectors to access a user's work data, such as emails and calendars, with permission. Because these connectors rely on the same underlying infrastructure as Outlook, they ran into errors when the Microsoft 365 services went offline.

OpenAI reported that new connections and some specific functions might fail for ChatGPT users during the incident. Similarly, Anthropic noted that most requests involving mail, calendar, and search through its connector were failing. These issues also affected Microsoft's own AI assistant, Copilot, which struggled to access files and work emails while the servers were struggling to respond to requests.

Microsoft Teams experienced a more nuanced set of problems during the outage. While the main chat and online meeting functions remained largely usable, features that require Exchange Online were impacted. For example, a user might have been able to stay in a meeting but would find it difficult to schedule a new one or open their calendar. This happened because the scheduling tools rely on the same backend service that was failing for Outlook.

Context and recovery

This is not the first time Microsoft has faced issues with these specific services. In January 2023, both Outlook and Teams went down together due to a global network change that interfered with connections to Microsoft 365 and the Azure cloud service. A similar failure in October 2025 disrupted the service because of a problem with Azure Front Door, which is the system that directs internet requests to the correct servers.

In the current incident, Microsoft has already begun restoring some capacity in certain regions, such as India. However, as of early Friday morning, a full recovery for all users remained unconfirmed. The company continues to monitor the situation to see if the network rollback and the new fixes hold steady as more traffic flows back into the system.

For organizations, the recovery process involves verifying that new messages are arriving and that calendar invitations are being processed correctly. Microsoft has provided specific incident numbers, such as MO1550821 for the wider incident and EX1550543 for the mailbox issue, so that administrators can track the progress. The final determination on the cause will depend on whether the current network adjustments successfully stabilize the server load.