Microsoft ‘Deeply Sorry’ After Global Outage Hits Outlook and Teams

Microsoft apologised on Friday after a fault dating back to an earlier network upgrade disrupted its services around the world. The company said a subsequent change had brought the problem to light. By early Friday evening in Britain, Microsoft said it had fixed the underlying fault and was still restoring service for some customers.
A Jury Sent Home and Bail Paid in Person
At Manchester Crown Court, Judge Maurice Greene sent jurors home after the outage disrupted the transport of prisoners and brought a trial to a halt. Someone, he told the jury, had been looking for a fax machine.
In Colorado, staff at the Arapahoe County jail relied on Outlook notifications to process online bail payments. When those messages stopped arriving, the sheriff’s office told anyone posting bail for a detainee’s immediate release to pay in person at the jail.
Maine’s state courts had their email back by about 7am local time on Friday.
They still could not say whether applications for protection orders emailed after 12.30pm the previous day had reached court staff.
The state’s judicial branch asked people who had applied during that period to consider sending their paperwork again or filing a paper copy. In Britain, several police forces asked people to phone with urgent matters after running into problems receiving crime reports online.
Some NHS staff went back to paper patient records.
In Essex, health officials opened a temporary phone line so care homes and other providers could make referrals or raise urgent queries about patients. Pharmacies had trouble reaching GP surgeries through NHS email, prompting the National Pharmacy Association to warn that some patients might have to wait longer for prescriptions.
University College London Hospitals said staff were temporarily unable to send or receive email and asked people with urgent enquiries to call their clinic or the switchboard. In Portsmouth, the hospital trust said the main problems were with staff email and administration. It told patients to attend planned appointments unless their care team contacted them.
Two removal crews in Kent had loaded their vans and were waiting for the keys to their customers’ new homes. David Strank, who runs the removals firm, told the Guardian that a third customer had called about a move booked for 23 November. Their solicitors were due to exchange contracts on Friday. They could not send emails.
At a financial firm in Taiwan, traders used a fax machine to confirm purchases and sales of short-term bills, Reuters reported. Foreign banks also had difficulty sending information to their overseas headquarters. In Europe, stock exchange operator Euronext asked listed companies to phone and had staff verify callers’ identities before handling matters usually dealt with by email.
Aviva had quarterly results to present on Friday. The British insurer sent out alternative dial-in numbers and used Bluetooth to transfer recordings. Its chief executive, Amanda Blanc, told Reuters that the company had tested its response to an outage of comparable scale about two weeks earlier.
How the Outage Began
Microsoft’s service health dashboard put the start of the outage at 4.17pm GMT on Thursday, 12 November, according to CIO.
The wave of public complaints came later.
By about 6pm, reports about Outlook had peaked at nearly 140,000 worldwide on Downdetector. The site recorded problems in 74 countries and territories within about an hour of the first wave of reports. Microsoft had yet to publish a total for the number of people affected.
Microsoft initially reported that some servers had unexpectedly become unavailable. That left fewer machines to handle incoming requests. The strain came from the loss of capacity, the company said, rather than a surge in demand.
In a statement reported at 5.37pm GMT on Friday, Microsoft said the fault had been introduced during an earlier upgrade intended to increase network capacity.
A subsequent network change exposed it.
The company did not specify which part of the earlier upgrade was faulty or explain precisely how the later change brought servers down. It ruled out a cyberattack and said its AI systems were unrelated to the failure. “We are deeply sorry for the disruption this incident has caused our customers and partners,” it said.
Much of the disruption centred on Exchange Online, the Microsoft 365 service that handles email and calendars.
Some Outlook users could not open their mailboxes in a browser or through the desktop app. Teams was more uneven. A call could go ahead while the calendar invitation for the next one failed. Copilot also had trouble retrieving information from some work emails and calendars.
OpenAI and Anthropic reported problems with the tools that connect their own AI services to Microsoft 365.
Why the First Repairs Failed
Some of the servers engineers brought back online stopped responding again once they began receiving requests. Microsoft limited the traffic reaching affected servers in some locations and reverted some machines to an earlier configuration. Engineers then brought traffic back in stages to see whether the repaired systems would stay up under load.
The company also tested a change to the automated checks that decide whether a server is ready to accept requests. Those checks normally steer traffic away from machines that cannot handle it. Mahmoud Elkhodr, a senior lecturer at CQUniversity in Australia, said a fault in the checks could take healthy servers out of service and put more pressure on the rest. His explanation drew on Microsoft’s public updates. The company’s evening statement did not identify the checks as the source of the original fault.
The network research firm Infoblox detected signs of significant recovery at about 11.20am GMT on Friday. Microsoft later reported sustained improvement across much of Europe, broader recovery in North America and progress in the Asia-Pacific region. By Friday afternoon, queues of undelivered email were close to normal. Users who could open their inboxes again might still be waiting for messages sent during the outage.
How Governments Fared
Norway reported disruption across several ministries. Greece’s digital governance minister, Dimitrios Papastergiou, said ministries using Outlook, Teams and Copilot had experienced problems but were still able to work. He told Greece’s Skai television that governments should develop alternatives for essential services.
Germany’s digital ministry said businesses had been widely affected, while the federal government reported no major disruption.
A ministry spokesperson told the Associated Press that the federal administration did not use Microsoft’s cloud services and its email software had remained stable. In Singapore, government agencies warned that replies to emails could take longer and directed urgent enquiries to call centres. Emergency services and Singpass, the national digital identity system, remained available.
The EU’s Data Act requires cloud providers to remove barriers to switching suppliers or using several at once. From 12 January 2027, providers will no longer be able to charge customers for switching, including for transferring their data out.
Sources: Associated Press, Reuters, Euronews, The Guardian
