Planning: DCL EPSCOR: SaTC Frontier: Ensuring Security and Privacy in AI-Enabled Cyber-Physical Systems

Artificial intelligence (AI) is increasingly used in cyber-physical systems (CPSs) that directly sense and interact with the environment to help them react to large, real-time data that comes in a variety of formats. However, the security and safety of AI models, and the privacy of the data used to build them, can be attacked. There is some research on AI model safety in other domains, but in CPSs the nature and scale of attacks on AI models may change because of their connection to the wider environment. This project will support planning a large-scale proposal to increase the security and privacy of AI-enabled CPSs. This will involve developing foundational knowledge and systematic tools to understand and defend against the unique risks of AI-enabled CPSs. It will also involve creating an engineering and education community capable of using that knowledge and those tools to build safer, more secure systems that help society better-interact with the world. The planning project is led by an interdisciplinary team with expertise in AI, machine learning, and CPS security. To better understand the challenges across a variety of CPS contexts, and provide resources for both expertise and deployment, the research team will collaborate with experts from a number of CPS domains, including space science, healthcare, transportation, and water resource management. The team will also work with stakeholders in education, business, the wider academic community, and government to inform questions and activities related to the project's education and workforce development goals. Planning activities include a series of biweekly seminars, quarterly newsletters, and symposia; these will support frequent communication and coordination around developing both the proposal itself and team required to execute it. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.