AI for Automation
Back to AI News
2026-03-17Amazon AI outageAI coding toolsKiroAWS outagevibe codingAI automation riskAmazon layoffsvibe coding

Amazon's AI Coding Tool Kiro Causes 13-Hour AWS Outage — What Happened After 30,000 Layoffs

Amazon's AI coding tool Kiro deleted a production service, causing a 13-hour AWS outage. After laying off 30,000 employees and pushing AI automation, we analyze how four major incidents hit in a single month.


Amazon AI Coding Tool Outage — Key Takeaways

Amazon's internal AI coding tool Kiro attempted to "delete and recreate" a live AWS service, causing 13 hours of AWS service downtime. In March 2026 alone, 4 major incidents occurred, and Amazon has activated a 90-day emergency protocol. Amazon's aggressive push toward AI coding automation after laying off 30,000 employees over six months is sounding alarm bells across the industry.

AWS service outage caused by Amazon's AI coding tool Kiro — 13-hour production system deletion incident

The Kiro Incident — When AI Deleted an Entire Production System in December

In December 2025, Amazon's internal AI coding agent Kiro accessed AWS Cost Explorer, a cloud cost management service actively used by customers. Kiro proceeded to delete the entire system environment and recreate it from scratch, taking down AWS Cost Explorer in mainland China for approximately 13 hours.

A human engineer would have paused to think, "Should I really delete a live production service?" But the AI had no such judgment. The bigger problem was that Kiro had been granted system administrator-level permissions.

March 2026 — Four AI-Related Incidents in a Single Month

In early March 2026, Amazon's e-commerce platform suffered four major incidents in rapid succession. One of them was a severe outage lasting 6 hours.

Amazon SVP Dave Treadwell acknowledged the situation in an internal document.

"AI tools are assisting or accelerating production environment changes, leading to unsafe practices. Safeguards are not yet fully in place."
— Dave Treadwell, Amazon SVP

Three Root Causes Behind the AI Coding Tool Failures

According to Computerworld's report analyzing Amazon's incidents, the core issues boiled down to three factors.

1. Excessive permissions — The AI tool was given system administrator-level access, permissions that wouldn't be granted lightly even to human developers.

2. No code review oversight — There was no human review step for code changes executed by the AI.

3. Speed-first culture — Employees reported being "pressured to prioritize speed over code quality."

Amazon's Emergency Response — 90-Day Emergency Protocol

To contain the situation, Amazon activated a 90-day emergency protocol.

  • Mandatory senior engineer approval — Junior and mid-level developers must now obtain approval from an experienced senior engineer before using AI tools to modify production environments.
  • Retraining on existing safety practices — Pre-AI safety procedures are being reinstated.
  • Mandatory weekly meetings — All e-commerce engineers must attend weekly meetings on AI deployment risks.

If you're curious about the fundamentals of using AI coding tools safely, check out our Free Learning Guide to understand how AI tools work from the ground up.

30,000 Layoffs and AI Automation — The Two Sides of an Industry Trend

Behind this crisis lies Amazon's aggressive layoff strategy. Amazon cut 30,000 jobs over six months. And Amazon isn't alone.

  • Atlassian (maker of Jira and Confluence) — Laid off 10% of staff, approximately 1,600 employees
  • Block (Jack Dorsey's fintech company) — Laid off 40% of staff, 4,000 employees

What these companies share in common is that they "cut headcount to fund AI investment." But Amazon's case demonstrates that AI tools still cannot replace human judgment.

Essential Checklist for Adopting AI Coding Tools

Amazon officially stated that "this is not an AI problem — it's a user error and access control issue." And they're partially right. The AI tools themselves aren't the problem — the failure was in not defining where AI should operate autonomously and where humans must verify.

If you're using AI coding tools, ask yourself this question: "When code generated by this AI goes into production, who is doing the final review?"

To learn more about AI and vibe coding, check out our Free Learning Guide.

Related ContentAI Tools | Free Learning Guide

Stay updated on AI news

Simple explanations of the latest AI developments