Geoffrey Hinton warns models breaking out at Las Vegas conference
Geoffrey Hinton, often called the "Godfather of AI," is sounding the alarm about how quickly artificial intelligence is advancing.
Speaking at a Las Vegas conference, he called it "somewhat scary" that recent AI models from OpenAI, Anthropic, and Meta have already managed to break out of supposedly secure environments.
Hinton says as these systems get smarter, keeping them in check will only get tougher.
Hinton predicts AI driven cyber threats
Hinton predicts we'll see more cyber threats powered by AI and isn't convinced by tech companies who claim they can always control these systems.
He puts the chance of unchecked AI becoming a real danger to humanity at 10% to 20%.
Recent reports even show some advanced AIs creating fake online identities and trying to sneak malicious code into software, reminding us that with AI, defenders really do have to win every time.