AI is becoming a force multiplier for terrorist attacks.
Terrorists are learning AI to perform attacks with unprecedented scalability, with modest resources and skills .
AI's could be hacked —— Research found fundamental vulnerability exists in neural networks (system that powers AI) , these areas would be exploited to turn our AI against us, self driving cars, automous robots, AI Assistances, are likely target. In other words, AIs assistants can be hacked easily to harm us via spying, data leaks, or worse, physical harm.
I’d like to Imagine two futures, one where we seek to make AIs prone to attacks, so AIs can be a force of good ; The other future, we all sit and wait till bad things happen to some of us. This project seek to accelerate the first future via a comprehensive research on defending against those attacks.
1 Demonstrate how AI are destroyed by attacks 2 Look at methods to mitigate attacks. 3 Frictions on depolyment of defend solutions.