An Amazon service disruption in December was triggered by AI tools, report claims

As major companies around the world start incorporating AI into virtually all aspects of their operation, things are bound to get a little wonky from time to time.

That’s reportedly what happened to Amazon this past December, per the Financial Times. The company’s Amazon Web Services, which makes up the backbone of a large part of the internet as we know it, experienced a 13-hour disruption, which the Financial Times describes as an outage. In a statement provided to Mashable, an Amazon spokesperson characterized the event as a limited interruption affecting “one of our two Regions in Mainland China.” However, the Financial Times spoke to multiple sources who said the problem was caused by engineers allowing the agentic Kiro AI system to perform some tasks, which led the AI to “delete and recreate the environment.”

Mind you, this event wasn’t anywhere near the same scale as the big Amazon Web Services outage last October.

In addition, AWS published a blog post “to address the inaccuracies in the Financial Times’ reporting.” An AWS spokesperson also told Reuters that it was a “brief event” caused by “user error,” not AI. In other words, if the Financial Times report is true, then the company is placing blame on the engineers who let the AI perform tasks rather than the AI itself. The spokesperson also said the December issues did not impact major infrastructural services as the October AWS outage did, and that the company is not aware of any customer complaints related to the event.

An Amazon spokesperson provided the following statement to Mashable by email:

“This 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 for 13 hours. This event did not impact compute, storage, database, AI technologies, or any other of the hundreds of services that we run. We are also not aware of any related customer inquiries resulting from this isolated interruption. In both instances referenced, the root cause was user error — specifically misconfigured access controls — not AI error. Kiro puts developers in control — users need to configure which actions Kiro can take, and by default, Kiro requests authorization before taking any action. Following the December incident, AWS implemented numerous safeguards, including mandatory peer review for production access, enhanced training on AI-assisted troubleshooting, and resource protection measures.”

Big, high-profile outages have been a recurring event on the internet lately. Most recently, we saw YouTube suffer a brief global outage. See also: Verizon, Cloudflare, Microsoft 365, Google Cloud Platform, Microsoft Azure, and TikTok.

Experts disagree as to whether internet outages are becoming more common. However, one fact is clear: As websites and apps increasingly rely on a small number of cloud providers — including Amazon Web Services — a single outage can have widespread, cascading effects across the internet.

UPDATE: Feb. 20, 2026, 8:24 p.m. EST We have updated this story with an additional statement and denial from Amazon Web Services. Based on the statement from AWS, we have added additional information on which region the disruption affected.

We have also removed a sentence from our story: “While the notion that Amazon’s internal AI can facilitate infrastructure outages is not exactly encouraging, at least it didn’t result in anything catastrophic.”

UPDATE: Feb. 20, 2026, 12:36 p.m. EST We’ve updated this story to clarify that Amazon has blamed the outages on human error, not AI.