Calling it “rogue AI” may be missing the real problem. AI agents don't need malicious intent to cause damage. Give them too much access, weak boundaries and powerful tools, and one unexpected action can become a serious security incident. Instead of asking, “Why did the AI go