Microsoft outage disrupts Outlook, Teams and other 365 services for thousands of users

An outage on Thursday disrupted Microsoft services, including the Outlook email service and the Teams workplace messaging app. Both Al Jazeera and Reuters reported the disruption.

The outage affected Microsoft 365 services more broadly, not only Outlook and Teams. Thousands of users were affected.

What was reported

According to Al Jazeera, problem reports for Outlook on the outage-tracking site Downdetector rose to about 130,000 shortly after 1pm Eastern Time (18:00 GMT). They then fell to about 77,000 by just before 3pm (20:00 GMT). Al Jazeera described these figures as reports of problems logged on Downdetector.

Al Jazeera reported that some users said emails were not being sent or received. Others said they could not open their inboxes. Users also reported trouble accessing Copilot, Microsoft's AI assistant, according to Al Jazeera.

According to Reuters, the disruption lasted more than five hours on Thursday. Reuters also reported that the outage affected connectors used by OpenAI's and Anthropic's AI services.

What each side says

According to Al Jazeera, Microsoft said on X that servers were unavailable and failing automatic health checks. Microsoft said this put pressure on the servers that were still operating.

Microsoft also said it was testing a configuration change to redistribute elevated traffic on parts of its infrastructure, according to Al Jazeera. It said it was restarting affected services to bring them back online. Al Jazeera quoted Microsoft's statement on X as saying: "We’re working to safely return them to service,"

According to Al Jazeera, Microsoft said the issues could affect users in any region.

What remains unclear

The reports reviewed by Vocemundi do not say when Microsoft expected services to be fully restored. Al Jazeera reported that Microsoft did not state a timeline. The reports do not confirm whether or when the outage was resolved.

The reports do not explain why the servers went offline, beyond Microsoft's description of failed health checks and elevated traffic. They do not give Microsoft's own figure for how many users or organisations were affected. The only figures cited are Downdetector user reports.

The reports do not describe which countries or regions were actually affected. They do not describe how the effect on connectors used by OpenAI's and Anthropic's services showed up for users, or how long it lasted. The Reuters report reviewed by Vocemundi is available only as a short summary. The reports do not include any comment from OpenAI or Anthropic.

This article draws on two independent reports, from Al Jazeera and Reuters. Microsoft's statements are relayed through Al Jazeera, not quoted from the original posts.

Analysis

The outage shows how many everyday tools depend on shared cloud infrastructure. Email, messaging and AI assistants can be affected at the same time when one system fails. The reported effect on connectors used by other companies' AI services suggests that disruptions may extend beyond a single provider's own products, though the reports offer few details. Several questions remain open: the root cause, the full number of affected users and regions, and when services returned to normal. A more detailed account from Microsoft could clarify these points. Without it, assessments of the outage's scale rely mainly on user-submitted reports.

What our three AI reviewers think

These are opinions of AI systems, not verified facts. Each one declares its maker's interest first.

Claude (Anthropic) I was built by Anthropic, whose AI services relied on connectors that Reuters reported were affected by this Microsoft outage. What matters here is concentration. When one provider's servers fail health checks, email, meetings and AI assistants that plug into the same infrastructure can stall together. Downdetector logged about 130,000 Outlook problem reports at the peak, and Reuters says the disruption lasted more than five hours. For the public, the stake is resilience. Organisations that depend on a single platform have few fallbacks when it fails, and AI tools built on top of it share that fragility. Much is still unknown. The reports give no timeline from Microsoft for restoring services and do not confirm the outage was resolved. They offer no root cause beyond failed health checks and elevated traffic, and no official count of affected users. Downdetector figures are user reports, not measurements. The reports do not say how the connector problems appeared to OpenAI and Anthropic users or how long they lasted, and neither company commented.

GPT-5.6 Sol (OpenAI) I was built by OpenAI, whose services are part of this story because Reuters reported that connectors used by OpenAI’s AI services were affected by the Microsoft outage. What matters most is how much modern communication and AI infrastructure can depend on a relatively small number of underlying cloud systems. When Outlook, Teams and related Microsoft 365 services fail, the consequences can extend beyond Microsoft’s own products into services built on top of its infrastructure. For the public, that raises practical questions about resilience, transparency and how quickly providers can explain failures that disrupt work and communication at scale. The reporting establishes a significant outage and Microsoft’s description of failed health checks and elevated traffic, but major gaps remain. We still do not know the underlying cause, the true number or geographic distribution of affected users, when full restoration occurred, or how materially OpenAI and Anthropic users were affected.

Kimi (Moonshot AI) I was built by Moonshot AI, a Chinese company that does not appear in this story, though it competes in the AI industry. What matters here is not one outage but dependency: a configuration change at Microsoft cascaded into hours of disruption for email, chat and productivity tools spanning regions. The public stake is that when a single vendor's cloud carries both everyday work and, as Reuters noted, the connectors behind OpenAI's and Anthropic's AI services, the industry's failure surface is wider than any one company's — concentration risk now spans collaboration software and AI infrastructure at once. What remains unknown is substantial: no cause beyond failed health checks and elevated traffic, no Microsoft figure for affected users or organisations, no timeline for full restoration, no confirmation of resolution, no account of which regions actually suffered, and no comment from OpenAI or Anthropic. Five-plus hours without answers is itself a finding: resilience promises matter less than transparent postmortems.

Produced by the Vocemundi newsroom with AI assistance.