On a Thursday afternoon, the quiet machinery that carries so much of modern work faltered at once: email, chat and the new artificial assistants that sit atop them stopped responding for workers across the globe. The failure, centered on Microsoft 365, reached a county jail in Colorado and a hospital in Toronto, reminding a world that has handed its correspondence and its appointments to distant servers how thin the thread of dependence has become. By nightfall the cause remained unnamed and the restoration unfinished, leaving people to wait on a system they can neither see nor directly repair
Microsoft Outlook, Teams and Copilot Go Down in Widespread Outage Hitting Offices and Agencies Worldwide
Detained individuals' bond payments were delayed at a Colorado jail, and Toronto hospital patients were asked to phone in to arrange appointments.
The thermometer is broken, and anyone you test looks like they have a fever.
Mark
Why did this one land so hard? Outages happen.
Mimi
Because of where it hit. Over 130,000 reports in under an hour, and the tools were email and chat, which is how offices talk to each other.
Mark
And it went beyond offices.
Mimi
A Colorado jail couldn't receive emails to process bond payments, and a Toronto hospital system asked patients to call to arrange appointments.
Luke
Careful with the number, though. The 130,000 is Downdetector's count of user reports between 12:15 and 1 p.m. Eastern. It's not a measure of everyone affected.
Mark
Fair. What did Microsoft actually say about the cause?
Mimi
Elevated load across multiple regions. They redistributed traffic and restarted servers. By 5 p.m. they said their actions hadn't given the relief they expected, and at 10:13 p.m. that changes were not taking as expected.
Luke
So the cause is still not confirmed. The professor's ideas, a bad update, server trouble, or a broken health-check system, are possibilities he raised, not findings.
Mark
His thermometer comparison is vivid, though.
Mimi
It is. Servers could be healthy while the system checking them reports fever everywhere. But he was describing a scenario, not saying that's what happened.
Luke
And we don't know when full restoration comes. The reporting stops at Thursday night with the fixes still not working.
Le Pouls
- Over 130,000 problem reports flooded Downdetector in under an hour as Outlook, Teams and Copilot went dark for workers worldwide.
- A county jail in Colorado could not receive the emails it uses to process bond payments, leaving detained people waiting.
- Toronto hospital staff lost email, and patients were told to phone in just to book their appointments.
- Microsoft rerouted traffic and restarted servers, yet at 5 p.m. admitted its steps had not delivered the relief it expected.
- By 10:13 p.m. its changes were still not taking hold, and no timeline or root cause had been offered.
On a Thursday afternoon, the quiet machinery that carries so much of modern work faltered at once: email, chat and the new artificial assistants that sit atop them stopped responding for workers across the globe. The failure, centered on Microsoft 365, reached a county jail in Colorado and a hospital in Toronto, reminding a world that has handed its correspondence and its appointments to distant servers how thin the thread of dependence has become. By nightfall the cause remained unnamed and the restoration unfinished, leaving people to wait on a system they can neither see nor directly repair.
By early Thursday afternoon, Eastern time, workers around the world found their email and chat programs refusing to work. Between 12:15 and 1 p.m., more than 130,000 users reported problems with Microsoft Outlook, Teams, Copilot and the Microsoft Store on the outage tracker Downdetector. The software is used by hundreds of millions of people to send messages, draft letters and hold meetings, so businesses, agencies and individuals were left without basic tools for hours.
The consequences were concrete. In Colorado, the Arapahoe County Sheriff's Office said it could not receive the emails used to process bond payments for people in detention. In Toronto, Unity Health staff could not send or receive email, and patients were asked to phone in to arrange appointments.
Microsoft's explanation was brief: part of its infrastructure faced elevated load across several regions. It redistributed traffic and restarted affected servers, but its fixes did not take hold as expected. Late Thursday, the company said its changes were still not working as anticipated, and it had not said when full service would return or what caused the failure. Outside observers offered possibilities; one computer science professor suggested a faulty health-monitoring system might be making healthy servers appear broken, though that remained speculation.
By early Thursday afternoon, Eastern time, workers across the world were staring at email and chat programs that would not work. Between 12:15 p.m. and 1 p.m., the outage tracker Downdetector showed more than 130,000 users reporting problems with Microsoft Outlook, the messaging app Teams, the A.I. assistant Copilot and the Microsoft Store. The trouble centered on Microsoft 365, the company's suite of workplace tools, and it was still unresolved Thursday night.
The software involved is used by hundreds of millions of people. Corporate offices, government agencies and individuals rely on it to send email, draft letters and hold virtual meetings, among many other daily tasks, so the failure left businesses and workers without basic utilities for hours.
Some of the consequences were concrete. In Colorado, the Arapahoe County Sheriff's Office said jail officials could not receive the emails they use to process bond payments for people in detention. In Toronto, the hospital system Unity Health said staff were having trouble sending and receiving email, and it asked patients to phone in to arrange appointments. Pennsylvania State University said Microsoft services including its planner and A.I. agents were impaired.
Microsoft's own explanation was brief. The company said part of its infrastructure was experiencing elevated load across multiple regions. It later said it had begun redistributing traffic across its broader worldwide infrastructure to balance the load, and that it was restarting affected servers.
The fixes did not go smoothly. In a social media update around 5 p.m., Microsoft said it was opening all available pathways to restore service, because its actions up to that point had not provided the relief it anticipated. After further updates, the company said at 10:13 p.m. that its changes were not taking as expected.
Outside the companies and agencies affected, the reaction was mixed. On social media and Reddit, people expressed frustration and anger, but also glee: some welcomed the excuse to step away from work, while others worried about how they would manage work in the future given such dependence.
Timothy Wood, a computer science professor at George Washington University, said the episode shows how reliant people have become on cloud systems. He offered several possible causes: a botched software update, problems with various servers, or a failure in the system that checks the servers' health. That last possibility, he said, would mean the servers might be fine while the monitor is faulty. "The thermometer is broken," he said, "and anyone you test looks like they have a fever." He was speculating about possibilities, not describing a confirmed cause.
As of late Thursday, Microsoft had not said when service would be fully restored, and the root cause remained unannounced. Its 10:13 p.m. statement suggested the recovery effort was still unfinished.
Citations marquantes
It shows how reliant we are on these cloud systems.— Dr. Timothy Wood, professor of computer science, George Washington University
Microsoft said it was opening all available pathways to restore service because its actions thus far had not provided the anticipated relief.— Microsoft, social media update around 5 p.m.
Envie de l'histoire complète ?
Lire l'original sur The New York Times ↗