Amazon calls engineers for a “deep dive” internal meeting to discuss “GenAI”-related outages

A normally optional, routine Amazon meeting turned serious when the brand’s top retail executive summoned engineers to discuss a series of AI-related outages.

While the memo leading up to the retail technology meeting, first reported by the Financial Times, didn’t specify which incidents would be on the docket for discussion, Amazon’s engineers have their pick of problems to sort through. In just one week, the e-commerce juggernaut experienced four incidents — each with a “high blast radius,” as described in the memo.

Per an Amazon spokesperson, the company’s “This Week in Stores Tech” (or TWiST) meeting is a weekly occurrence and a “normal part of business” where “a specific group of retail technology leaders and teams…review operational performance across our team store.”

That recurring session was upended on Tuesday to do a deep-dive on what’s been behind the recent outages.

And it seems AI-assisted production changes are to blame.

A deep dive into deeply troubling AI-assisted coding

“The availability of the site and related infrastructure has not been good recently,” David Treadwell, SVP ecommerce services at Amazon, writes in a note to employees, as reported by CNBC.

Based on the evidence, Treadwell might have been putting it lightly. Last week, both Amazon’s website and app faced severe outages; for six hours, customers could neither check out, access account information, nor view product prices. At the time, Amazon released a statement saying the issues were related to “software code deployment.”

Now the brand’s internal memo reveals the recent incidents were tied to “GenAI-assisted changes.”
Another memo from Treadwell shows that these AI-assisted coding errors have been creating problems for Amazon as far back as Q3 2025.

He calls out GenAI tools used to supplement or speed up production changes for “leading to unsafe practices,” noting that “best practices and safeguards” for the use of generative AI haven’t yet been concretized.

What safeguards are in place, Amazon plans to “reinforce.” Namely, senior engineers will be called upon to review the GenAI-assisted production changes prepared by junior employees.

Treadwell also proclaims: “We are implementing temporary safety practices which will introduce controlled friction to changes in the most important parts of the Retail experience, in parallel we will invest in more durable solutions including both deterministic and agentic safeguards.”

More incidents on the heels of recent AWS outages

Amazon’s AI-assisted slip-ups come at a time when money seems to be pouring into infrastructures — and away from employees.

Like other hyperscalers, Amazon is going hard on infrastructure spending, with its February 2026 earnings report announcing an expected $200 billion investment in capital expenditures for the year. At the same time, Amazon is still tightening the purse strings on headcount. The company has laid off tens of thousands of workers since 2022, with the last round in January 2026 affecting 16,000 corporate workers.

In the midst of website and app issues, Amazon’s cloud computing division is also facing turbulence.

The end of 2025 saw several Amazon Web Services outages. One incident in particular, reported by the Financial Times, caused 13 hours of downtime for a cost management feature — a brutal side effect of Kiro-led changes —an agentic IDE that aims to help engineers automate or accelerate code changes.

An Amazon spokesperson said the AWS outages are not related to this month’s website and app incidents, but it continues a worrying trend of AI-assisted mistakes.