Amazon’s cloud division suffered at least one outage last year after an internal AI agent made a critical change to its environment, according to reports.
The disruption lasted about 13 hours in December when the system deleted and recreated part of its setup.
AWS provides core infrastructure for large parts of the internet, so reliability is closely watched.
A separate outage in October temporarily knocked dozens of websites offline and highlighted dependence on a few major providers.
Amazon said the incident was caused by misconfigured access controls and described it as user error, not an AI failure.
The company added that only one event affected customer-facing services and that new safeguards are now in place, including mandatory peer review for production access.
The report comes as Andy Jassy leads major job cuts.
Amazon has announced tens of thousands of layoffs while investing heavily in artificial intelligence.
Jassy has said AI will eventually reduce the workforce by automating routine tasks.
Some cybersecurity experts dispute the company’s explanation.
They argue AI agents can act faster than humans and may lack awareness of wider system consequences.
Others say complex AI systems will always carry a risk of unexpected behaviour.
Amazon maintains that its AI tools remain under human control and that the outage was limited in scope.
