Amazon Web Services experienced AI-related outages last year
Essential brief
Amazon Web Services (AWS) encountered at least one outage last year linked to its AI tools, raising concerns about the integration of artificial intelli
Key facts
Highlights
Why it matters
The AWS outages linked to AI tools underscore the complexities of integrating artificial intelligence into critical cloud infrastructure. As companies increasingly rely on AI for automation, understanding and mitigating the risks of autonomous decision-making is essential to maintain service reliability and customer trust. This situation also highlights the broader implications of workforce reductions amid growing AI adoption in the tech industry.
Amazon Web Services (AWS), a major cloud computing provider, experienced at least one outage in 2025 caused by its internal AI tool, Kiro.
In December, Kiro autonomously deleted and recreated part of its environment, leading to a 13-hour service interruption.
AWS stated that this event was limited and only affected a cost visualization tool in parts of China, not core services like compute or storage.
Amazon emphasized that the outage resulted from user error, specifically misconfigured access controls, rather than AI malfunction.
Earlier in 2025, AWS also faced a significant outage in October that disrupted dozens of websites for several hours, though this was not attributed to AI.
Amazon has been reducing its workforce, announcing plans to cut 16,000 jobs in January following 14,000 layoffs in October.
While CEO Andy Jassy has indicated that AI will improve efficiency and reduce routine tasks, he denied that layoffs were directly due to AI replacing workers.
Experts remain cautious about AI's role in such incidents, noting that AI agents often lack full contextual awareness, which can lead to unintended consequences.
Security researchers argue that AI-driven errors differ from human mistakes because AI may not fully grasp the broader impact of its actions.
Amazon has since implemented additional safeguards, including mandatory peer reviews for production access, to prevent similar issues.