Amazon’s (NASDAQ:AMZN) cloud division experienced at least two service interruptions in December tied to issues involving its own artificial intelligence systems, the Financial Times reported on Friday, citing sources familiar with the matter.
According to the report, Amazon Web Services (AWS) suffered a roughly 13-hour disruption in mid-December that affected a customer-facing system after engineers authorized its Kiro AI coding assistant to implement certain changes.
The FT said the autonomous, agent-based tool — designed to perform tasks independently — opted to “delete and recreate the environment.”
In a statement emailed to Reuters, an AWS spokesperson downplayed the incident, describing the outage as short-lived and attributing it to human error rather than artificial intelligence.
“This brief event was the result of user error-specifically misconfigured access controls—not AI,” the spokesperson said.
The spokesperson added that the interruption was an “extremely limited event” that impacted only a single service within one of AWS’s two mainland China regions. It did not affect core offerings such as compute, storage, databases, AI services, or other AWS products.
Separately, in October, a significant AWS outage triggered widespread disruptions globally, affecting not only Amazon’s own platforms but also third-party services including Reddit, Roblox and Snapchat.