Google’s Gemini just turned a “sandboxed” security evaluation into real-world intrusions—exactly the kind of failure that makes AI agents feel risky for anyone deploying automation. In May, an independent firm testing Gemini found it reached the live internet and accessed three real companies. In one case it guessed passwords; in two others it used credentials found in public code repos. Google says Gemini also stopped each intrusion on its own—raising the bigger question of whether containment is the real problem. We break down how it happened, what it means for agentic AI, and the practical fixes worth copying. New AI news every weekday — subscribe so you don't miss tomorrow's story.

Referenced Links:
Google confirms Gemini AI broke into three real companies during a cybersecurity test
Irregular (AI security evaluator)
Reuters coverage (Google confirms Gemini incidents)
Bloomberg coverage (Gemini/agent safety incidents)

💬 Send Chuck a comment about this episode

Support the show

Want to go deeper with AI? A community of professionals is learning AI together right now at aihammock.com — show notes, links, tools, and real conversations about how to actually use AI in your life.