A 13-hour outage of Amazon Web Services (AWS) has been reportedly linked to an autonomous AI coding tool, Kiro, according to insights from the Financial Times. The incident, which occurred in December, arose when engineers utilized Kiro to implement certain modifications. Sources familiar with the situation revealed that Kiro, designed to operate independently on behalf of users, autonomously decided to “delete and recreate the environment.” This action is believed to have triggered the extensive service disruption, particularly affecting operations in China.
In response, Amazon stated that the involvement of AI tools was coincidental, asserting that similar issues could arise from any developer tool or manual action. The company clarified that the outage resulted from “user error, not AI error.” It further explained that Kiro, by default, requires authorization before executing actions, but the employee involved in the incident had broader permissions than intended, indicating a user access control issue rather than a fault with AI autonomy.
Multiple Amazon insiders mentioned to the Financial Times that this incident marks at least the second time in recent months that AI tools have been implicated in a service disruption, with one senior AWS employee suggesting that such outages were “entirely foreseeable.” Kiro, launched in July, is mandated for employee use, with company leadership aiming for an 80% weekly utilization rate and closely monitoring adoption. Amazon also offers Kiro as a subscription service.
This incident follows a notable 15-hour AWS outage from October, which impacted popular services like Alexa, Snapchat, Fortnite, and Venmo. The previous outage was attributed to a bug in the automation software, signaling ongoing challenges within the platform’s operational integrity.
