Amazon is reportedly introducing new rules for how its engineers write, review and deploy code. These changes are reported to be a direct response to recent outages that have affected the company’s e-commerce operations in recent months, some of them linked to its own AI coding tools. Citing internal documents, Business Insider claims that the measures include a 90-day temporary safety reset that will be in addition to existing policies. The report further claims that “the new policy targets approximately 335 ‘Tier-1 systems,’ or services that can directly impact consumers, that have experienced multiple order-impacting incidents since last year and are owned by VP-level organizations.”
What are the new rules
The trigger for the reset is reportedly “several major” incidents that Amazon’s SVP of e-commerce services that Dave Treadwell, Amazon’s SVP of e-commerce services, described in an internal note to staff on Tuesday as a “trend of incidents”. The scale of some of those failures was significant, the report said.On March 2, customers across Amazon’s marketplaces were shown incorrect delivery times when adding items to their carts, the report pointed out, adding that this incident resulted in nearly 120,000 lost orders and around 1.6 million website errors. Amazon’s AI coding assistant Q was purportedly identified as one of the primary contributors.Three days later, on March 5, an outage caused a 99% drop in orders across Amazon’s North American marketplaces – leading to an estimated 6.3 million lost orders in a single day. The root cause, according to an internal document, was a production change deployed without going through Amazon’s formal documentation and approval process. Under the 90-day reset, Amazon engineers working on Tier-1 systems must reportedly now:Get two people to review their work before making any coding changes. This requirement had apparently lapsed or been bypassed in some teams.Use an internal documentation and approval tool called Modeled Change Management for all production changes.Use an automated coding system that strictly follows Amazon’s central reliability engineering standards.An Amazon spokesperson clarified to the publication that the policy does not require junior or mid-level engineers to get sign-off specifically from senior engineers for AI-assisted changes.Meanwhile, for directors and VP-level leaders, Amazon is said to be instructing who own Tier-1 systems to audit all production code change activities within their organisations — a top-down review of how code has been written, approved, and deployed.
