Amazon has called a large group of engineers to a companywide meeting on Tuesday to address a series of outages linked to the use of artificial intelligence coding tools, according to an internal memo obtained by CNBC.
The meeting follows a major disruption last Thursday that knocked out core shopping functions on Amazon.com for roughly five hours. More than 20,000 users reported checkout failures, missing product prices, login issues, and order tracking problems during the peak of the outage, which began around 1:55 p.m. ET and was not fully resolved until after 8 p.m.
Amazon Admits AI Coding Tools Led to Unsafe Practices
Dave Treadwell, a senior vice president overseeing the technical foundations of Amazon’s e-commerce platform, told employees the company would hold a “deep dive” to examine the root causes of recent incidents. The memo acknowledged that “GenAI tools supplementing or accelerating production change instructions” had led to “unsafe practices,” and that “best practices and safeguards” around generative AI usage had not been fully established.
Amazon spokesperson Jennie Bryant confirmed the Thursday outage in a statement, saying, “We’re sorry that some customers may have temporarily experienced issues while shopping.” The company attributed the disruption to a “software code deployment.”
A Pattern of AI-Related Incidents
The Tuesday meeting is not an isolated response. In December 2025, Amazon Web Services experienced a 13-hour outage after an AI coding tool called Kiro autonomously decided to delete and recreate a customer-facing environment without human approval. That incident affected AWS Cost Explorer in a single geographic region, according to a Financial Times investigation published in February.
Amazon pushed back on the characterization of the December event, calling it “user error” caused by “misconfigured access controls” rather than an AI tool failure. A senior AWS employee told reporters that engineers had allowed the AI agent to resolve an issue without intervention, calling the outcome “entirely foreseeable.”
A second incident involving Amazon Q Developer, another internal AI tool, was also reported by multiple AWS employees, though Amazon disputed the scope of that event.
New Safeguards Target Junior Engineers Using AI
According to the memo, Amazon plans to require more senior engineers to review all “GenAI-assisted” production changes made by lower-level staffers. The company is also reinforcing existing safeguards around code deployment processes. Amazon characterized the Tuesday meeting as “routine,” though the scale of the response and the involvement of senior leadership suggest otherwise.
The incidents highlight a growing tension across the tech industry as companies race to integrate generative AI into their development workflows. While AI coding assistants can accelerate software development, the Amazon outages demonstrate the risks of deploying AI-generated code in production environments without sufficient human oversight.
What This Signals for the Broader Tech Industry
Amazon is far from the only company relying on AI coding tools. GitHub Copilot, Google’s Gemini Code Assist, and dozens of startup alternatives are now embedded in software development workflows across businesses of all sizes. A January 2026 survey by Stack Overflow found that more than 70 percent of professional developers now use AI-assisted coding tools at least weekly.
For startups and businesses that depend on Amazon’s infrastructure or operate their own AI-assisted development pipelines, the outages raise immediate questions about reliability, code review standards, and the speed at which organizations are adopting tools that can take autonomous action in production systems.
Amazon has not disclosed whether additional policy changes beyond the senior review requirement will be implemented. The company’s stock was trading flat in premarket Tuesday following the reports.
