Amazon’s famous AWS outages shook the internet back in December 2025. There are now reports suggesting those were linked to the company’s own artificial intelligence tools. The report comes from Financial Times investigation.
The most significant disruption lasted 13 hours in mid-December after engineers deployed Kiro, an agentic AI coding tool, to fix a minor software bug. Instead, the autonomous tool determined it needed to “delete and recreate the environment,” disrupting the Cost Explorer system that AWS customers use to analyze service costs. Multiple Amazon employees also told the Financial Times that Amazon’s Q Developer product, an AI-enabled chatbot for code writing, had triggered an earlier outage affecting a different system.
Amazon disputed that its AI tools were responsible, characterizing the incident as a “user access control issue, not an AI autonomy issue.” The company noted that the staff member involved had “broader permissions than expected” and said Kiro requests authorization before taking action by default.
Company insider sources told Reuters that “the same issue could occur with any developer tool or manual action.” Amazon also emphasized that no customer-facing services were affected by the second incident and that neither approached the scale of an October 2025 AWS outage that disrupted OpenAI’s ChatGPT for 15 hours, among many other notable web-services.
Following the disruptions, Amazon says it implemented mandatory peer review and additional staff training. However, concerns remain among about the rapid deployment of AI agents, especially when they are given a level of freedom that can cost the entirety of the internet.
