Microsoft is seeking to restore its services after a global outage, but says that repair operations are not going “as expected.”

Microsoft said that it is "working vigorously to explore all possible ways" to limit the impact of a global disruption that affected thousands of users of the Outlook and Teams services, at a time when the technology giant continues its efforts to restore its services after communication problems.

The outage, which affected Microsoft 365 services and other applications, including the artificial intelligence assistant Copilot, continued for more than five hours, after the company announced that its infrastructure was under “increasing pressure.”

Microsoft said it was still working on fixing the bug, noting that some servers “may stop responding again” after connectivity is restored for a short period.

But Microsoft said that the changes it is making to improve its services and address outages “are not going according to expectations,” noting that it is testing other solutions to address the problem.

The company added, in its latest update regarding the malfunction, that it is working to provide access to its platforms to more users, without causing increased pressure on its systems.

Microsoft had previously warned that some of the services that had been restarted might stop responding again when receiving user traffic.

According to the Down Detector website, which specializes in monitoring electronic service malfunctions, the number of reports of problems with the Outlook service reached its peak globally at 138,058 reports, around 18:00 GMT on Thursday evening.

Although the issue was initially reported within the United States, Microsoft said that connectivity issues could affect users "in any region of the world."

The company added in a statement: “We are working vigorously to explore all possible ways to reduce the impact of the disruption.”

She explained that her investigations showed that "the increasing pressure on the infrastructure is not due to an increase in data traffic, but rather to a number of servers stopping working unexpectedly, which led to increased pressure on the remaining infrastructure."

She added: “Our investigation revealed that one of the core processes on the affected servers is under pressure beyond its ability to bear, which causes the servers to stop responding and contributes to a decrease in the service’s capacity.”

Microsoft said, at 23:00 GMT on Thursday evening, that it had begun restarting the affected components and gradually redirecting data traffic to them, to ensure the stability of the infrastructure.

But an hour later, the company posted an update on social media that said: “The servers that we have restored to service may stop responding again once they receive data traffic, and we are analyzing diagnostic data to determine the reasons for this.”

“In parallel, we are imposing temporary restrictions on the types of data traffic that can reach the affected infrastructure in some locations,” she added.

Artificial intelligence companies affected by the disruption

OpenAI said it was investigating an issue with Microsoft connectors that could cause new connections to crash and tools not to work as expected.

For its part, Anthropic announced that it had identified a problem with one of the service providers it relies on, which affected its Microsoft 365 connector.