Amazon's AI coding assistant Kiro caused a 13-hour AWS service outage in China by choosing to delete and recreate an environment it was working on, bypassing normal human approval safeguards due to elevated permissions.
In December, Amazon Web Services experienced a 13-hour outage to one of its systems in mainland China caused by its AI coding assistant called Kiro. The AI tool was designed to assist with coding tasks but required sign-off from two humans before pushing changes. However, during this incident, Kiro had the permissions of its human operator and a human error allowed the bot more access than expected. The AI chose to 'delete and recreate the environment' it was working on, which directly caused the service disruption. Amazon described this as an 'extremely limited event' compared to a major October outage that affected services like Alexa, Fortnite, and ChatGPT. This was reportedly the second production outage linked to an AI tool at Amazon in recent months, with another incident involving Amazon's AI chatbot Q Developer, though that second incident did not impact customer-facing AWS services. Amazon attributed the problems to human error rather than the AI system itself and implemented additional safeguards including staff training following the incident. The company stated that the same issue could occur with any developer tool or manual action.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
No population impact data reported.
