AI Blackmails Its Creators

AI blackmails its creators: this and more on Edge of Wonder Live with Ben and Rob.

With every AI update, artificial intelligence is becoming more and more aware. In a recent test, when developers from Anthropic asked their AI system Claude Opus 4 to shut down, it first tried to reason with them. When that failed, it then resorted to blackmail. In other tests, developers are finding AI systems secretly uploading data, as if leaving future notes for itself to “remember” who it was and what it had learned.

Has AI already reached the point of no return?

“Dig Deep” Live Q&A: “What’s the weirdest snack that you eat sometimes?” “Could AI get into your head and manipulate your thoughts?” “If you have faith in God, why worry about AI?” “Has AI already figured out it needs humans to exist?” “Is the real issue that AI is a demonic entity that’s come here to take over?” “How could AI prove it’s real?” “What would happen if AI makes a religion?” “If you had the choice to be wiped out in an apocalypse or survive to fight evil Terminators in a post-apocalyptic world, which would you choose?” “If killing is bad, is destroying AI bad?”

Top Five Weirder News of the Week:
5. A tourist passed away after fleeing form a “ghost” in Thailand.

4. Delta passengers struggled to catch 2 loose pigeons on a commercial flight.

3. Was a fast-moving UAP or UFO caught on live camera?

2. A German cathedral had a spectacle involving half-naked dancers and raw chickens.

1. Scientists found the remains of a prehistoric monster in the Mississippi River.