Microsoft 365 had two separate Exchange Online problems on 29 October 2026, one hitting encrypted email replies and the other hitting Copilot Chat's conversation history. Both were Microsoft's fault at the infrastructure level, and the Copilot Chat issue ran for 151h 53m before Microsoft called it largely resolved.
Started
16:00 UTC
Duration
Lasted 1d 6h
Source
IsDown
Next time
Microsoft 365 is degraded. Is your product working?
An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.
50 monitors for your own websites. No card.
Affected components: Exchange Online
What happened?
Microsoft logged two incidents under Exchange Online starting on 29 October 2026. The first began at 16:00 UTC: external users' replies to encrypted email messages were not being delivered. Microsoft's 21:59 update said an anti-spam rule was being triggered on internal domains tied to encrypted message delivery, and they applied an exclusion to those domains to stop the block. That incident ran for 30h 28m before resolution. The second began at 17:19 UTC and affected Microsoft 365 Copilot Chat: users were sporadically missing recent portions of their conversation history, though Microsoft was clear the affected conversations were not deleted, just temporarily unavailable. Microsoft's updates describe requests from the same user being intermittently routed to different parts of their infrastructure, so a conversation sometimes loaded an earlier version missing the latest exchanges. Over the following updates Microsoft said they traced the issue to specific geographic regions and eventually to resource limitations, applying a "secondary fix" across affected regions by 00:38 UTC on a later date. That second incident lasted 151h 53m in total, the longest of the day by far. Microsoft published 15 status updates across both incidents, and no user reports were filed into this window on the record. Microsoft's history shows 59 incidents in the last 90 days with a median duration of 9 hours 3 minutes, so a multi-day incident like the Copilot Chat one is well outside their typical pattern.
Learning
This day shows two different failure shapes under the same Exchange Online umbrella: a misfiring anti-spam rule blocking a narrow category of mail, and a routing problem scattering a user's own requests across backend infrastructure so their chat history looked incomplete. Neither of these would show up cleanly on an external uptime check, because the service kept responding and authenticating, it just served the wrong data or silently dropped a specific reply path. That is the gap between synthetic monitoring and a real outage: your probe hits an endpoint and gets a 200, while a real user behind a specific domain or region gets broken behavior that never touches your own infrastructure. UptimeRobot's 30 day data for Microsoft 365 recorded 26 incidents, all resolved, with a median duration of 13h 22m, which lines up with Microsoft's own pattern of multi-hour, occasionally multi-day fixes. If your product depends on Exchange Online delivery or Copilot Chat, watching your own endpoints only tells you your half is fine. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.