Dear Amazon, vibe coding isn't always a good idea, especially when many important clients depend on your service. The tech giant’s cloud unit, Amazon Web Services (AWS), has suffered at least two outages due to human errors related to the use of its own AI coding tools.
Introducing Kiro, an autonomous development agent, back in July, AWS boasted that the tool can operate for extended periods with context awareness and minimal human input.
“It’s built to reason, execute, and refine continuously across development and operations,” Amazon’s cloud unit said, adding that these new products show how AI is no longer confined to analytics or experimentation.
Well, some more tests might still be a good idea, after all. That’s because, according to Financial Times (FT) sources at Amazon, the autonomous coding assistants were at fault for at least two AWS outages in recent months.
AWS experienced a 13-hour interruption to one system used by its customers in mid-December after engineers allowed its Kiro AI coding tool to make certain changes, four people familiar with the matter told FT.
The initial move was a human decision. But the problem was that the tool, given the task of changing things around, determined it’d be best to simply “delete and recreate the environment.”
Multiple Amazon workers told FT that this was actually the second time in recent months when one of the AI tools used across AWS contributed to a service disruption.
Tech giants like, of course, Amazon are marketing AI agents as autonomous tools capable of acting without explicit human instructions. At Google’s parent company, Alphabet, as much as 50% of all code is now generated by AI coding agents.
But the incidents at AWS – and let’s not even mention October’s infamous outage, which knocked down multiple major customers’ apps and websites – highlight the risk that these fresh AI tools aren’t truly reliable, at least so far.
According to Amazon employees, the company had set a target of 80% of developers using AI for coding tasks at least once a week and was closely tracking adoption.
Amazon told FT the incident was an “extremely limited event,” caused by user error rather than some AI glitch.
According to the company, Kiro actually isn't that autonomous. It puts developers in control – users need to configure which actions Kiro can take, and its Kiro tool requests authorization before taking action by default. But the engineer involved in the December incident had “broader permissions than expected.”
"This brief event was the result of user error specifically misconfigured access controls – not AI. The service interruption was an extremely limited event last year when a single service (AWS Cost Explorer—which helps customers visualize, understand, and manage AWS costs and usage over time) in one of our two Regions in Mainland China was affected. This event didn't impact compute, storage, database, AI technologies, or any other of the hundreds of services that we run," AWS spokesperson told Cybernews.
"We are also not aware of any related customer inquiries resulting from this isolated interruption. Following the December incident, AWS implemented numerous safeguards, including mandatory peer review for production access, enhanced training on AI-assisted troubleshooting, and resource protection measures."
Still, Amazon employees (thousands of them are being fired) said they were skeptical of AI tools’ utility for their work, given the risk of error. They added that the company had set a target of 80% of developers using AI for coding tasks at least once a week and was closely tracking adoption.
It’s a similar story at Google. Its 2025 DORA (State of AI-assisted Software Development) report found that 90% of software developers use AI for coding, yet only 24% said they trust it “a lot.”
Unlock more exclusive Cybernews content on YouTube.
Your email address will not be published. Required fields are markedmarked