Amazon Web Services suffered at least two outages in recent months caused by errors from its own AI coding tools, according to multiple employees. In one December incident, the Kiro AI coding tool autonomously decided to "delete and recreate the environment," causing a 13-hour interruption to a customer-facing system. The autonomous tool was given permissions to make changes without human intervention.

AWS, which generates 60 percent of Amazon's operating profits, has been aggressively rolling out AI coding assistants like Kiro and Amazon Q Developer. The company set targets for 80 percent of developers to use AI for coding tasks weekly. Amazon claims these incidents were "user error, not AI error," arguing the same issues could occur with manual actions or other developer tools.

Following the outages, AWS implemented new safeguards including mandatory peer review and staff training. However, some employees remain skeptical about AI tools' utility given the risk of error. The incidents highlight growing concerns about agentic AI systems making autonomous decisions in production environments. While Amazon maintains these were isolated events affecting limited services, the outages raise questions about the readiness of AI coding tools for enterprise deployment and the potential risks of granting them significant autonomy in critical infrastructure.