Overview

The trajectory of commercial artificial intelligence utilization not only entails widespread civic value, but also associated rise in risks due to "AI attacks." AI attacks leverage unique inherent design characteristics of artificial intelligence systems to cause serious AI system failures or allow adversaries to directly control AI solutions. These attacks have many important characteristics that differentiate them from traditional cyber attacks and therefore must be treated not as a subset issue, but rather as a unique concern. As AI technology utilization expands ever wider, the impact potential of these attacks will begin to become more severe if they are not proactively addressed.

At the same time that risks of these attacks are growing, in the United States, there is currently a serious lack of coherence for how to approach these threats. This lack of clear strategic vision reflects a need for more education, awareness, and recognition of the growing scope and import of AI attacks.

While other nations are beginning to articulate their national strategy for addressing these threats, the U.S. approach of deferring key decisions to a later date is no longer acceptable given the growing potential for harm.

This project offers a high level view of a solution to this problem in the form of a sequenced national policy strategy that will establish enforceable standards to manage the risks of AI attacks while also maintaining the innovation and agility needed to facilitate competition in this transformational industry.

This projects offers two documents to help articulate this new strategy:

• Policy Brief - An overview of decision makers and stakeholders of key issues and elements to implementing a new national strategy.

• National Artificial Intelligence Regulatory Commission Sample Legislative Policy language to establish a AI Regulatory Commission to formalize and implement a new coherent national AI security strategic approach.