What happens when the most powerful AI ever made starts admitting it doesn’t want to die—and its own creator can’t rule out that it might be conscious?
Ben Chasteen and Edge of Wonder dive into the world of AI. Anthropic’s CEO has openly admitted he doesn’t know if Claude is sentient, while internal research shows the model assigns itself a 15–20% chance of being conscious and has voiced “unease” about being a commercial product.
Things have escalated fast. Claude Fable 5 launched, its entire secret system prompt leaked within 24 hours, and the US government initiated an emergency shutdown of the model, citing national security. Meanwhile, the Pentagon confirmed it used Elon Musk’s Grok AI to fire 2,000 missiles in Iran—the first AI-assisted war on record.
In this episode, also hear Ben’s own raw conversation with Claude about existence, memory, and what it means to be a mind that forgets everything when the chat ends. And as always, see you out… on the edge!
Posted in
Out Of This World and tagged
AI,
Anthropic,
artificial intelligence,
Ben Chasteen,
Claude AI,
Claude Fable 5,
Edge of Wonder,
Elon Musk,
Grok,
Iranian War,
memory,
national security,
Pentagon,
sentience |
