Amazon Web Servies is denying a report that alleges an AWS outage in December was caused by its own Kiro AI tool making the error.
Amazon is denying a report that an AWS outage in December was caused by AWS engineers who allowed its own Kiro AI coding product to conduct changes, which spurred the cloud outage.
“This brief event was the result of user error—specifically misconfigured access controls—not AI,” said an AWS spokesperson in a statement to CRN.
A recent report from the Financial Times said December’s cloud outage stemmed from errors involving AWS’ own AI tools, citing sources familiar with the incident. Specifically, AWS’ agentic Kiro AI coding tool decided to autonomously “delete and re-create the environment,” which spurred the AWS outage in late 2025, according to the report.
[Related: Andy Jassy On AWS’ $244B Backlog, Trainium4 And AI Chips Strategy]
AWS denied that report, saying the outage was due to user errors based around misconfiguration of access controls. December’s outage took down sites and services, including its own Amazon Alexa, Ring and Prime Video.
AWS said it was a coincidence that AI tools were involved in December’s outage, and “the same issue could occur with any developer tool or manual action.”
“The service interruption was an extremely limited event last year when a single service (AWS Cost Explorer—which helps customers visualize, understand, and manage AWS costs and usage over time) in one of our two Regions in Mainland China was affected,” Amazon said.
Amazon said its Kiro AI tool “requests authorization before taking action” by default, so AI did not bypass human engineers.
“Kiro puts developers in control—users need to configure which actions Kiro can take, and by default, Kiro requests authorization before taking any action,” AWS told CRN.
Launched in July, Kiro is an agentic coding service that works alongside users to turn prompts into detailed specs, then into working code, documents and tests. Kiro’s agents aim to help customers solve problems and automate tasks like generating documentation and unit tests.
AWS Implements ‘Numerous Additional Safeguards’
In a statement to CRN, AWS said the company has implemented numerous safeguards after December’s cloud outage.
“We implemented numerous additional safeguards, including mandatory peer review for production access,” Amazon said.
“This event didn’t impact compute, storage, database, AI technologies, or any other of the hundreds of services that we run,” Amazon added.
AWS generated $35.6 billion in total sales during the fourth quarter of 2025, up 24 percent year over year.
The Seattle-based cloud company’s annual run rate is now $142 billion.
