Two Amazon cloud outages in December triggered by AI tools

As major companies around the world begin to integrate AI into virtually every aspect of their operations, things are likely to get a little weird from time to time.

This is what happened to Amazon last December, according to the Financial Times. The company’s cloud infrastructure, which forms the backbone of much of the Internet as we know it, experienced two minor outages that month, including a 13-hour outage in the middle of the month. This was apparently caused by engineers allowing the Kiro AI agentic system to perform certain tasks, leading the AI ​​to “delete and recreate the environment.”

Please note, this event was not on the same scale as the major Amazon Web Services outage last October. An AWS spokesperson told Reuters it was a “brief event” caused by “user error”, not the AI ​​itself. In other words, if the latest report is true, then the company is placing the blame on the engineers who let the AI ​​perform tasks rather than the AI ​​itself. Regardless, the spokesperson also said the December outages did not impact key infrastructure services like the October one did.

While the idea that Amazon’s internal AI can facilitate infrastructure outages isn’t exactly encouraging, at least it hasn’t resulted in anything catastrophic.

Large, high-profile outages are a recurring occurrence 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 on whether internet outages are becoming more common. However, one fact is clear: As websites and applications increasingly rely on a small number of cloud providers, including Amazon Web Services, a single outage can have widespread cascading effects across the Internet.