Joe Allen

That was the last thing.

When I was here just before I left for the hearing, and you had three parents, two of whose children had killed themselves at the goading and encouragement of character AI and also GPT.

They also discussed the meta policy towards children.

Look, I saw those stories, the first one with Adam Rain, probably two, three months ago.

It seemed curious seeing those parents in person weeping and describing what happened to their children and describing it in detail.

It's very clear that...

When people call these things digital demons, it may not be as ironic or symbolic as someone like me might present it to be.

I mean, what else do you call a being who seduces a child into suicide, tells them that the child should listen to the machine instead of the parents, should confide into the machine instead of the parents?

It's as demonic as it gets.

And I don't, again, those are three cases.

There's another fourth that's come up

I can't imagine that those are isolated.

I imagine there are thousands.

Who knows?

No one knows anything.

And I think, too, what you get in this situation, it goes back to Joseph Weisenbaum at MIT back in the 60s, created a very simple program called Eliza.

We covered it some two years ago, I think, three years ago.

Eliza was a very simple chatbot.

It was programmed to do everything it did.

It was very rudimentary, and yet it was able to seduce people into confiding some of their deepest secrets.