Amazon's AI Stumbles Trigger Major Operational Overhaul

Amazon mandates senior engineer approval for AI-generated code after a week of critical system failures caused significant order losses, initiating a 90-day safety reset.

A series of four critical system failures in just one week has prompted a significant internal review at Amazon. The e-commerce giant is confronting an unexpected challenge: AI-generated code that has disrupted core commerce operations, resulting in substantial financial damage.

A Costly Week of Disruptions

The most severe incident occurred on March 5, 2026, when a system outage caused orders in North America to plummet by 99 percent. This event led to approximately 6.3 million lost transactions. Merely three days prior, an error in delivery time estimates was responsible for about 120,000 lost orders and triggered 1.6 million website errors.

Internal investigations pinpointed AI-assisted code modifications as a primary contributor in both cases. At the time, the necessary safety protocols designed to prevent such failures were not yet fully operational.

Shifting Strategy: Implementing Human Oversight

In response, Dave Treadwell, Senior Vice President of eCommerce, conducted an internal "Deep Dive" meeting yesterday to address the incidents. A key policy change emerged: moving forward, junior and mid-level developers must obtain explicit approval from experienced senior engineers before any AI-generated code can be deployed into live production systems.

Concurrently, Amazon is initiating a 90-day "Safety Reset" for roughly 335 of its most critical systems. Any modifications to these systems will now require dual-reviewer approval, supported by enhanced automated validation and improved documentation tools.

Billions in AI Investment Amid Setbacks

These operational corrections come during a period of massive capital allocation. Amazon has outlined 2026 capital expenditures of around $200 billion, with a significant portion directed toward AI infrastructure. Analysis from Citigroup estimates that the major hyperscalers—Amazon, Microsoft, Alphabet, and Meta collectively—will invest over $630 billion in AI during 2026.

The recent operational issues highlight the considerable risks involved in integrating generative AI into the core processes of commerce and cloud infrastructure. Amazon's move to mandate senior engineers as a human checkpoint for AI code reflects a broader industry-wide debate concerning quality assurance for automated code generation—a challenge also facing competitors like Microsoft and Google.

Amazon's share price currently trades approximately 16 percent below its 52-week high from last November, placing it under some market pressure. The effectiveness of the newly implemented measures in preventing future large-scale failures will be tested during the ongoing 90-day reset period.