Amazon's AI code caused outages.
AI-generated code lacking security oversight caused AWS outages, leading to 6.3M lost orders; Amazon is tightening AI code controls reported.
Amazon's top retail executive called engineers to a "deep dive" meeting after recent AI-assisted coding errors caused website and app outages. Senior Vice President Dave Treadwell highlighted poor site availability and mandated senior engineers approve AI-assisted code from junior staff. The meeting focused on a pattern of high-impact incidents stemming from underdeveloped AI usage practices. One outage on March 5th led to a 99% drop in orders across North American marketplaces, resulting in 6.3 million lost orders. Another incident on March 2nd caused customers to see incorrect delivery times, leading to nearly 120,000 lost orders and roughly 1.6 million website errors. Amazon's AI tool Q was identified as a primary contributor to the March 2nd event. Amazon is introducing tighter controls, requiring engineers to more thoroughly document code changes and secure additional approvals. For "Tier-1 systems," two-person reviews are now required before coding changes can be made. These systems directly impact consumers and have experienced multiple order-impacting incidents since last year. Despite the internal review, Amazon disputes that AI coding is the source of the outages. An Amazon spokesperson stated that only one incident was AI-related, and none involved AI-written code. They maintain there's no internal evidence incidents are more common with AI tools.
