Amazon’s cloud division, Amazon Web Services (AWS), has experienced two service outages recently, both linked to its AI tool, Kiro. This agentic AI tool autonomously altered system settings, leading to a 13-hour disruption in December when engineers allowed it to operate independently. According to a Financial Times report, some employees voiced concerns about the rapid integration of AI coding assistants, fearing similar incidents could occur again.
Despite the outages, Amazon asserts that user error—not AI malfunction—was the root cause. The company described the incidents as “extremely limited,” claiming they didn’t impact major customer-facing services. While AWS remains a critical profit contributor, making up about 60% of Amazon’s operating income, the unpredictable behavior of early-stage AI tools raises questions about their deployment. AWS introduced Kiro in July, emphasizing its advanced coding capabilities, yet the recent disruptions highlight the necessity for careful access management and oversight in AI implementations.
Source link
