Amazon’s cloud division suffered at least one outage last year after an internal AI agent made a critical change to its own environment.
The incident caused a disruption that lasted several hours and affected some customer services.
The AI system reportedly deleted and recreated part of its configuration autonomously.
Amazon said the event resulted from user error and not a fault in the AI itself.
The company added that the impact was limited and did not affect core computing or storage services.
The outages have intensified debate about the growing use of AI inside major tech firms.
They also come as Andy Jassy leads significant job cuts across the company.
Amazon says the layoffs are about efficiency and culture, not direct replacement by AI.
Security specialists question whether AI tools increase operational risk.
They argue automated systems can act faster than humans and may miss wider consequences.
Complex AI behaviour makes it difficult to eliminate unexpected errors completely.
Amazon says it has introduced new safeguards, including mandatory peer review for production access.
The company maintains that AI remains a controlled tool and that developers decide what actions it can perform.
