Amazon’s cloud division suffered at least one outage last year involving an internal AI system, according to reports.
An incident in December lasted about 13 hours after an AI agent automatically deleted and recreated part of its environment.
The disruption affected only a limited number of services, the company said.
Amazon argued the problem was caused by misconfigured access controls and described it as user error rather than an AI failure.
AWS underpins large parts of the internet and many public-sector systems.
A separate outage in October briefly knocked dozens of websites offline and renewed concerns about reliance on a few major providers.
The reports come as Andy Jassy oversees major job cuts at Amazon.
The company announced 16,000 layoffs in January after 14,000 roles were removed earlier.
Jassy has said AI will improve efficiency but is not the direct reason for the reductions.
Some experts dispute Amazon’s explanation.
They argue AI systems can act faster than humans and may lack awareness of wider consequences.
That can increase the risk of mistakes if controls are not strict.
Amazon says it has added new safeguards, including mandatory peer review for production access.
It maintains that developers remain responsible for what the AI tools are allowed to do.
