Amazon Clarifies AI Outage Reports: Only One Incident Involved AI Tools, No Code Issues
Key Takeaways
- ▸Only one of multiple recent Amazon.com outages involved AI tools; the root cause was an engineer following inaccurate AI-generated guidance from an outdated wiki, not AI-written code
- ▸Amazon clarified that none of the outages were caused by AI-written code, contradicting widespread media reporting based on the original Financial Times article
- ▸The incidents were limited to Amazon's retail infrastructure and did not involve AWS services or introduce new engineer approval requirements for AI tool usage
Summary
Amazon has issued a correction to recent Financial Times reporting that falsely attributed multiple service outages on Amazon.com to AI-written code. In a detailed statement, Amazon clarified that only one of several recent incidents involved AI tools in any way, and in that case the root cause was not AI itself but rather an engineer following inaccurate guidance from an AI tool that had referenced outdated internal documentation. The company emphasized that none of the outages were caused by AI-written code, and that the incidents were separate, unrelated operational issues affecting Amazon's retail infrastructure.
The incident in question involved an engineer receiving incorrect advice from an AI-assisted tool that had inferred information from an outdated internal wiki. Amazon stated it has since addressed the issue and updated internal guidance to prevent similar problems. The company also refuted claims that AWS services were involved in the outages or that Amazon has implemented new approval requirements for engineers using AI tools. The correction comes after the Financial Times' initial report spawned additional coverage of the false claims across other media outlets.
- Amazon has updated internal guidance and addressed the underlying issue to prevent similar incidents from occurring in the future
