A Hardware "Stop Button" for AI Applications

Motivation

Navigating the rise of Artificial Intelligence (AI) is like steering a ship into uncharted waters. There's a lot of potential out there, but also unknown risks. As AI becomes more capable and more unpredictable, we probably need a "big red button" that only we as humans can press when things go wrong. But how? Here's the core of what we're thinking about:

Our proposal is based on the following assumptions:

Therefore, building hardware-embedded safety precautions seems to be a great approach to make them attack-resistant.

Desired Features of the Hardware "Stop Button"

What are the target use cases, and the key features of this hardware precaution?

Methodology

After analyzing the necessary features of the hardware stop button, we realized that we need to find a hardware feature that we can adjust to hinder AI performance but not non-AI applications. Here are some of the architecture features that we can tune:

Here are some applications that usually run on GPUs, but each of them are sensitive to different hardware resources:

After the sensitivity study, we found that the memory bandwidth is a unique limiting factor for AI applications. Therefore, the best design for the AI "stop botton" is probably to adjust the memory bandwidth.