Amazon is aggressively pushing its engineers to adopt AI tools. At least 80% of developers are expected to use AI for coding tasks at least once a week. However, recent events suggest that this accelerated rollout may have come at a cost.
As the Financial Times reported, Amazon Web Services suffered a 13-hour outage in December after engineers allowed AI coding tool Kiro to update code without oversight. Kiro decided that the best solution was to “delete the environment and recreate it.” This is probably one way to fix a problem.
This wasn’t an isolated case. A subsequent FT report revealed that Amazon’s e-commerce business has been struggling with a “trend of incidents” since the third quarter of 2025, prompting a company-wide deep-dive meeting led by SVP Dave Treadwell.
Some employees were already skeptical about whether these AI tools were actually useful in their day-to-day work, and these incidents didn’t exactly help build trust.
How bad has it gotten?
Business Insider obtained internal documents that paint a clearer picture of what actually happened. On March 2, 2026, Amazon’s AI coding tools contributed to an incident that caused 120,000 lost orders and 1.6 million website errors.
Three days later, on March 5, 2026, a separate outage caused a 99% decline in orders across North American marketplaces, resulting in 6.3 million lost orders. That’s a number that will certainly show up on the bottom line of a financial review, even for a company as large as Amazon.
What is Amazon doing to ensure something like this never happens again?
Amazon is currently rolling out a 90-day security reset for around 335 critical systems. Engineers must have two people review changes before deployment, use a formal documentation and approval process, and follow more stringent automated controls.
The company claims that these were user errors, not AI errors, and that the same errors could happen with any developer tool. That’s a fair point, but it doesn’t change the outcome.
When artificial intelligence tools are given broad permissions without proper oversight, things break, and the volume of code generated by AI only increases the damage.
