Microsoft’s precarious AI ecosystem laid bare in wake of major service outage

A massive outage at Microsoft has raised concerns in the firm’s enterprise customer base after a failure in Bing’s API disrupted Copilot and ChatGPT for several hours.

This outage has revealed the extent to which these products are precariously balanced on a single system which, in turn, many businesses are also reliant on to deploy their own suites of Microsoft AI tools.

Users reportedly began encountering error messages on May 23, with those attempting to use both Bing and Copilot greeted with messages such as “this page isn’t working right now” or “we’re unable to connect to this service”, respectively.

A failure in Bing’s API is understood to be the issue which, in addition to bringing down Bing itself, also had severe consequences for an entire ecosystem of products that depend on it.

ChatGPT’s internet capabilities were knocked out, with OpenAI describing the situation as a “partial outage” and preventing users from accessing the chatbot.

Microsoft, meanwhile, warned users they would be unable to access the Copilot service due to the fault.

Other search engines such as DuckDuckGo and Ecosia also went offline, owing to the fact that they too use Bing’s API and rely on that platform to deliver search results.

Some services are now back up and running, such as ChatGPT, though the service update page for Microsoft 365 still says that “users may be unable to access the Microsoft Copilot service”.

Bing API outage could raise enterprise concerns 

Jason Kent, Hacker in Residence at Cequence, told ITPro this outage should raise concerns due to the fact it highlights the interwoven nature of critical Microsoft services. 

“Today's outage on Bing's search results API demonstrates how a single point of failure can have widespread consequences,” he said.

“When Bing's API went down, services that depend on it, like Copilot, ChatGPT, and DuckDuckGo, were also unable to return search results.”

Kent noted that businesses need to ensure they pay closer attention to the services they are using in order to be prepared for issues of this kind.

“Organizations should know what APIs they have, what APIs they use, and what to do if a vendor’s APIs go offline for an extended period of time,” he said.

An increasing reliance on third-party vendors demands that organizations understand the systems that they are using, even if the problems that affect them are out of their hands.

“I cannot imagine a five-hour API outage that wouldn’t impact a single business let alone an interconnected business like Microsoft,” he added.