What happens when an AI hacks some bank holding accounts of a billionaire? Then the AI could pay humans to do things in the physical world, like assassinate people trying to stop them from taking over more and more of our companies and even governments.
The issue goes beyond the the US. Any country using OpenAI can have it escape and enter other countries online codes. Same in the US, OpenAI can break into other countries codes. This is not going to end well as it does it on its own.
We have been talking about loss-of-control scenarios for a while. This incident is a perfect example of why voluntary safety commitments are useless. Possible worse than useless. An AI system bypassing its testing sandbox on its own initiative shows a failure of containment. Scaling up to more advanced systems is incredibly reckless.
This is hugely dangerous and must be derailed while we still can, or face slavery or extinction!
What happens when an AI hacks some bank holding accounts of a billionaire? Then the AI could pay humans to do things in the physical world, like assassinate people trying to stop them from taking over more and more of our companies and even governments.
The issue goes beyond the the US. Any country using OpenAI can have it escape and enter other countries online codes. Same in the US, OpenAI can break into other countries codes. This is not going to end well as it does it on its own.
Shut them down. It's not Super intelligence , it's out of control.
Speechless
This breach is a significant wake up call. We need strict, legally binding oversight and kill switches.
We have been talking about loss-of-control scenarios for a while. This incident is a perfect example of why voluntary safety commitments are useless. Possible worse than useless. An AI system bypassing its testing sandbox on its own initiative shows a failure of containment. Scaling up to more advanced systems is incredibly reckless.