Unpopular opinion: The AWS outage wasn't the AI's fault.

Unpopular opinion: The AWS outage wasn't the AI's fault.

Amazon's AI coding assistant Kiro caused a 13-hour AWS outage, and the headlines are already asking: "AI agents go rogue, who's responsible?"

Amazon says it was "misconfigured access controls," not the AI. But analysts predict 40% of enterprise AI agents get cancelled this year due to governance gaps just like this.

Here's the uncomfortable truth: Amazon's not wrong. Bad permissions are the root cause. Whether human or AI clicks the button, if your production environment can be wiped by a single mistaken identity, you've got bigger problems.

But businesses keep making the same mistake, trying to bolt on governance after deployment.

The companies actually winning with AI agents? They locked down permissions, audit trails, and kill switches BEFORE turning anything loose.

Before you deploy an AI agent with production access, ask: can you prove who did what, when, and why? If not, fix that first. Guardrails aren't optional anymore.

What's the biggest/favorite mistake you've seen so far?