Abstract:
Microsoft Learn posted a post on social platform The post quickly became a forum for users to complain, receiving more than 1.2 million views and more than 1,100 replies; the most liked piece of advice was: "Make sure your product doesn't go down."
Other users also sarcastically said "Don't rely on Microsoft products", or reminded Microsoft not to deploy server changes hastily or put AI-generated code without sufficient inspection directly into production. Others said that Outlook, Teams and Copilot had problems at the same time, accidentally becoming a "free tutorial" for working without AI. "Microslop," a derision directed at Microsoft's AI products, also appeared in the comments.
Many replies linked the failure to Microsoft's personnel adjustments and asked the company not to lay off employees who know how to repair the system. At the end of October, Microsoft announced a voluntary severance buyout plan for eligible employees on some teams, covering employees level 70 and below in the United States, Canada, the United Kingdom and India. Windows Latest pointed out that this does not mean that Microsoft laid off 60,000 employees; about 60,000 are reported to be the number of people who may be eligible for the program, and the actual scope of participation varies depending on organizational and business needs.
Charles LaManna, Microsoft's executive vice president of Copilot, agents and platforms, later apologized for the glitch. Microsoft said at the time that some servers failed to pass automatic health checks, and servers that were restored became unresponsive after traffic picked up. Core processes on the affected servers were overwhelmed. The company is pausing deployments, rolling back recent network configuration changes, and gradually pushing fixes to upstream authentication and routing components; Microsoft said there is no indication that the outage was caused by a cyberattack.
Microsoft has not yet attributed the accident to AI-generated code or personnel adjustments, and a complete post-incident investigation report was not released at the time. The number of Outlook problem reports on the fault monitoring website Downdetector once reached about 130,000, but this is only the number of user reports and is not equivalent to the actual number of people affected. As of the time of the report, Microsoft's service status page still shows that multiple Microsoft 365 services are in a downgraded state.
Comments