This is the first part of a series about the Principle of Minimal Privilege, also known as the Principle of Least Authority.
One technique that is quickly trotted out in discussions about making AI safer is just not giving the AI agents access to certain things. A naive suggestion is not giving AI control over anything in the physical world. However, this has several problems, including:
- Control over a digital resource can often be used to obtain control over a physical resource
- An AI can pretend to be a human or passing information from a human
- A lot of humans are open to following instructions from AI
In computing, there are multiple ways to protect access to... (read 244 more words →)