Amazon’s cloud division reportedly suffered at least two outages last year linked to its own AI tools.
The incidents have raised questions about the company’s expanding use of artificial intelligence.
According to reports, a 13-hour disruption in December occurred after an AI agent called Kiro deleted and recreated part of its environment.
Another outage in October disrupted dozens of websites for several hours.
Only one of the AI-related incidents is said to have affected customer-facing services.
AWS provides critical infrastructure for much of the internet.
Its dominance has already sparked debate over reliance on a few major providers.
The company has also secured 189 UK government contracts worth £1.7bn since 2016.
Andy Jassy confirmed plans to cut 16,000 jobs in January, following earlier layoffs.
He has argued that AI improves efficiency, though critics question its growing operational role.
Amazon insists the outages resulted from user error rather than AI failure.
It says misconfigured access controls caused the disruption and that additional safeguards are now in place.
These include stricter reviews and limits on automated actions.
Some cybersecurity experts remain sceptical.
They warn that AI agents may lack awareness of wider system consequences.
They argue that preventing similar mistakes in complex environments may prove difficult.
