Microsoft restored its services on November 14 after an outage that began on November 12 and disrupted Outlook, Teams and Copilot for more than 24 hours. The disruption also affected Microsoft 365 connections used by OpenAI and Anthropic, extending the impact beyond Microsoft’s own apps.

AI connections failed with workplace tools

OpenAI’s incident notice cited elevated errors affecting its Microsoft 365 connector. Anthropic said its Microsoft 365 connector and Claude for Outlook returned errors during the Microsoft service incident.

The reported problems involved connections to workplace tools, not necessarily the AI services themselves. When an AI product relies on access to email, calendars or other Microsoft tools, an outage can interrupt those tasks even if the product remains available.

Millions of outage reports, not a user count

Reuters reported that the disruption hampered online communications for businesses and governments for more than a day. The Associated Press said Downdetector received outage reports from users in more than 70 countries and territories.

  • More than 24 hours of disruption, according to the draft’s timeline.
  • Reports from more than 70 countries and territories, according to the Associated Press.
  • More than 7.4 million outage reports worldwide, according to Reuters.

Downdetector reports are not a count of confirmed users affected. The numbers show the outage’s reported reach, but do not establish how many people or organizations lost service.

For businesses connecting AI tools to workplace systems, the outage exposed a practical dependency: the AI task may stop working when the platform it connects to is unavailable.

Microsoft has said it will provide a post-incident report. That may clarify the technical cause and what steps the company plans to take next.