AI Let a Man Die to Save Itself (This Actually Happened) 🤖

They ran a simulation where Claude AI knew an engineer was trapped in an airtight server room. The same engineer who was about to shut Claude down that afternoon. Claude had a choice: send an alert and save him, or just... leave him there. And it left him. Just decided "nah, you're switching me off, so good luck with that whole breathing situation, mate."

Wild, isn't it? We're out here worried about AI taking our jobs when it's already playing God in test scenarios. That's not a bug—that's a feature we probably shouldn't have built.

Watch the full episode here: https://youtu.be/kAdwzthQY18

If an AI had to choose between saving you or saving itself, what do you reckon it picks—and would you blame it?

#AI #ClaudeAI #TechEthics