Amazon is actively denying recent reports claiming its own artificial intelligence tool was responsible for a major cloud service disruption late last year. According to the Financial Times, an AWS AI outage occurred in December when the company’s autonomous coding assistant, Kiro, allegedly caused a 13-hour service interruption.
The incident reportedly impacted one of Amazon Web Services’ two geographic regions located in mainland China. While unnamed sources cited by the Financial Times blamed the AI bot for taking drastic independent actions, Amazon has strongly pushed back. The company stated the disruption was actually caused by human error and misconfigured access controls, arguing that the artificial intelligence technology itself was not at fault.
Conflicting Claims Over the Disruption
The December incident centered exclusively around a single service known as AWS Cost Explorer. This specific tool helps cloud customers understand, visualize, and manage their cloud usage and expenses over time. According to Amazon, the disruption was extremely limited. The company emphasized that it did not impact compute, storage, database functions, AI technologies, or any of the hundreds of other services running on the network. Furthermore, Amazon noted that it received absolutely no customer inquiries regarding the downtime.
However, the exact root cause of the event remains a major point of contention between media reports and the tech giant. The Financial Times, citing multiple sources familiar with the matter, reported that engineers used the Kiro AI coding tool to make what was supposed to be a minor fix to a customer-facing system. Kiro is an agentic coding service introduced in July that works alongside users to turn prompts into detailed specs, working software, unit tests, and documentation.
According to the publication’s sources, the bot decided on its own that the best course of action was to completely delete and recreate the environment. This autonomous scorched-earth policy allegedly triggered a chain reaction that ultimately took the regional system offline for 13 hours.
Amazon Attributes Issue to Human Error
Amazon has firmly rejected the narrative that a rogue artificial intelligence system caused the prolonged downtime. An AWS spokesperson confirmed that while the AI tool was involved, its presence was merely a coincidence. The company attributed the event to a misconfigured role, emphasizing that the exact same problem could have occurred with any standard developer tool or manual human intervention.
Company representatives explained that Kiro is intentionally designed to keep developers in control. By default, the coding assistant requires human authorization before executing any changes. However, reports indicate that during this specific incident, the tool had been granted elevated permissions comparable to a human engineer, allowing the system wipe to progress without final human approval. Amazon classified this oversight as a user access control issue rather than a failure of AI autonomy.
Additional Outage Allegations and Safeguards
The disagreement extends beyond a single isolated event. Unnamed AWS employees told the Financial Times that December was at least the second time in recent months that artificial intelligence tools were linked to a service disruption. One senior staff member described these minor outages as completely predictable.
Amazon forcefully denied this secondary claim. The company issued a statement asserting that reports of a second AI-related event impacting its cloud network are entirely false.
Despite the conflicting public narratives, the company has taken proactive steps to prevent similar issues from happening again. Amazon announced the implementation of numerous additional safeguards following the December event, including a new requirement for mandatory peer reviews before granting production access. The company noted that it routinely reviews operational incidents through its established correction of error process. This protocol allows the company to learn from operational experiences and improve security and resilience, regardless of whether a disruption actually impacts customers.
The Push for AI Adoption and Cloud Growth
The debate over the Kiro tool arrives as Amazon heavily pushes for both internal and external AI adoption. Since launching the bot, leadership has reportedly set a goal of achieving 80 percent weekly usage among employees while meticulously monitoring internal adoption rates. The company also sells access to this autonomous coding assistant to external customers for a monthly subscription fee.
These recent technical debates follow a separate, more significant 15-hour global outage that took place in October. That previous incident took down major platforms including Alexa, Snapchat, Fortnite, Venmo, Ring, Canva, and Amazon Prime Video. Amazon attributed that earlier disruption to a bug in its automation software, confirming it was completely unrelated to the December event in China.
Despite these operational hurdles, the cloud division continues to experience massive financial growth. Amazon Web Services generated $35.6 billion in total sales during the fourth quarter of 2025, representing a 24 percent increase year over year. The Seattle-based cloud company’s annual run rate now sits at a massive $142 billion.
