AWS responds after report claims cloud services outages sparked by use of internal AI tools—What the company said

Amazon Web Services has experienced at least two outages in the last few months due to its in-house AI coding assistants, according to a report by Financial Times. However, in a statement to Mint, AWS said that the claim that the outage was caused AI is false and the root cause of the issue was ‘user error’.

“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. Following these events, we implemented numerous additional safeguards, including mandatory peer review for production access. Kiro puts developers in control—users need to configure which actions Kiro can take, and by default, Kiro requests authorization before taking any action.” AWS said in a statement to Mint.

Original story follows below:

Reportedly, the disruptions have raised doubts about the tech giant’s ability to roll out these coding assistants. The company had suffered a 13-hour outage in December to a system used by its customers due to an engineer allowing Amazon’s Kiro AI assistant to make autonomous changes.

The report noted that the agentic tool determined that the best course of action was to “delete and recreate the environment”. Amazon is also said to have posted an internal postmortem of the ‘outage’ surrounding the AWS system.

Reportedly, it was the second time in recent months in which one of Amazon’s AI tools had been at the centre of a service disruption.

“We’ve already seen at least two production outages [in the past few months]… The engineers let the AI [agent] resolve an issue without intervention. The outages were small but entirely foreseeable,” FT quoted an AWS employee as saying.

AWS is said to be looking to build and deploy AI agents capable of taking actions independently on behalf of users. The company is said to be looking to sell this technology to outside customers as well.

Amazon reacts to AWS outage:

The tech giant told FT that it was “coincidence that AI tools were involved” and that “the same issue could occur with any developer tool or manual action”.

“In both instances, this was user error, not AI error,” Amazon added while stating that it had not seen evidence that mistakes were more common with AI tools.

It also added that the December incident was an “extremely limited event” which only affected a single service in parts of mainland China. Amazon added that the second incident did not have an impact on an “extremely limited” event.

The FT report, while citing Amazon employees, says that the group’s AI tools were treated as an extension of an operator and were given the same permissions. In both the cases, the engineers involved were not required to get a second person’s approval before making changes, as would have normally been the case.

Meanwhile, Amazon responded by saying that its Kiro tool “requests authorisation before taking any action” but the engineer involved in the December incident had “broader permissions than expected — a user access control issue, not an AI autonomy issue”.

Some Amazon employees have reportedly said that they are sceptical of the AI tools’ utility for the bulk of their work given the risk of error. The employees also noted that the company had set a target of 80 percent for developers to use AI for coding-related tasks at least once a week and was closely tracking adoption.