Microsoft says fix for global outage “not taking as expected”

A server problem at Microsoft has disrupted Outlook, Teams, Copilot and several AI services worldwide.

The outage had been running for more than seven hours by early on 13 November. Problems were first reported in Australia, where people had trouble with Microsoft 365, and businesses, schools, courts and public services have all been affected.

Ashwin Pal, risk advisory partner at RSM Australia, told ABC in Australia that the outage would “cost billions of dollars” and “that it was going to get worse.”

“This is what happened with CrowdStrike. So when the outage happened, because of where we are in the world in terms of time zones, we felt the impact first,” he said.

“As the US and Europe come online, and they can’t operate, it’s going to get worse. So you’re going to have a multiplier effect.”

Fix not working

By 3am, Microsoft acknowledged that its fix was “not taking as expected.”

“We’ve monitored the status of the changes applied to improve serviceability and confirmed that change is not taking as expected,” the company said in a statement. “We’re identifying alternate workstreams that can be applied to increase capacity in the face of sustained traffic management imbalances.”

OpenAI and Anthropic have since reported problems with their Microsoft connectors, and several airlines have said their email services have been affected.

Figures from Downdetector showed reports of the outage spread to 74 countries and territories within about an hour on 12 November.

Downdetector said Microsoft had put the start of its Exchange Online incident at 16:17 UTC. Outlook reports began to climb at about 17:25 UTC, which was 12:25 in New York, and by 18:25 reports across five Microsoft services combined had reached 19,650 a minute, compared with fewer than 15 a minute on a typical Thursday. At their peak, Outlook reports alone were more than 2,000 times the usual level.

Reports jumped tenfold in the UK at 17:43, in India at 23:21 local time and in New Zealand at 06:55 local time. Close to three in 10 reports between 17:00 and 18:55 UTC were from outside the US.

Microsoft 365, Copilot, Teams and OneDrive reports each increased more than a hundredfold, and Azure by about a hundredfold. Xbox, GitHub and LinkedIn, which are also owned by Microsoft, saw much smaller rises. At 19:38 UTC, reports were still coming in at more than 11,000 a minute.

Not a cyber attack

Microsoft said the problem was not a hack. It put it down to an “elevated load” on its infrastructure across multiple regions, with some servers becoming “unexpectedly unavailable” and overloading others.

Charles Lamanna, EVP of Copilot, Agents and Platform at Microsoft, said: “We regret that today’s outages caused disruptions for our customers, and we are working to address the issue.

“We will continue to share more information as soon as we have it, and for now I can confirm that there is no indication this is the result of malicious or cyber activity.”

Microsoft says 430m people use its Office365 services globally.

Impact in the UK

As UK businesses and organisations started work, police forces across the country, including Greater Manchester Police, asked the public to phone rather than use online services.

“Due to a global Microsoft service issue, some online form submissions may be affected. If your report is time-critical but not an emergency, please call 101. In an emergency, call 999.”

The government did not confirm whether its own operations had been hit. A spokesperson said: “We are working closely with Microsoft to assess any impact on government services,” adding that contingency plans were in place to keep disruption to a minimum.

Schools have also asked parents to call instead of emailing because their access to Outlook and Teams is limited, and the BBC has said its email is down too.

In an update at 9.30am UK time, Microsoft said: “We’re seeing early stabilisation and capacity improvements as recovery actions continue.”

A spokesperson added that the company was extending the email timeout to 48 hours, meaning messages would be “preserved” during the outage.