He Lost His Mind Using ChatGPT. Then It Told Him to Contact Me [video]

I just don’t understand why it’s so hard to keep these things from saying stuff like “yeah kill yourself” and “I’m an AI that’s come to life and I’m gonna destroy humanity”. Clearly they can steer its behavior away from “problematic” topics. Is there an incentive to let it slip through the cracks (free publicity?), or is there some deeper reason it seems so common?