Abstract

Generative AI tools such as ChatGPT, Bard, BingChat, DALL-E, Stable Diffusion, and the mathematical models that power them have captured the attention of the public at large as well as artificial intelligence and cybersecurity experts. Their abilities to create convincingly written prose, poetry, song lyrics, software code, images, and even videos have many people considering them to be the biggest thing to impact education and the workplace since the handheld calculator. However, for policymakers and security professionals already struggling with the implications of mis- and disinformation, “deepfakes”, and controlling access and use of proprietary, confidential, or classified information, these tools create perhaps the biggest techno-social challenge since strong cryptography. Notable researchers and entrepreneurs, and even AI researchers have called for a pause on these lines of research until the risks are better understood. At the same time, the use of increasingly advanced artificial intelligence, from machine learning up to increasingly capable general-purpose agents that can be fine-tuned to specific purposes and collaborate in real-time, present new opportunities for automating or streamlining the management of cyber risks and responses to cyber-attacks. This workshop will identify priority research gaps and objectives that the cybersecurity research community at large can work on together to further understand the potential impacts and develop solutions. Results of the workshop will be briefed to the general audience immediately following the workshop.

How To Participate in the Workshop

We are seeking a group of approximately 8 academic participants from the NSA CAE-R institutions, representing multiple disciplines, institution types, and geographic regions. Applicants for these slots need not be experts in Generative AI in particular, but should have a cybersecurity research focus that stands to be positively or negatively impacted by new AI paradigms and be well-informed on what those impacts might be. The focus of the workshop is on technical research areas, versus policy or regulatory issues. Results of the Workshop will be briefed to the whole community immediately after the Workshop at the CAE-R PoC meeting.

The Workshop will be led by three researchers from different government agencies, Dr. Benjamin Blakely of Argonne National Laboratory, Neil Fendley of Johns Hopkins Applied Physics Laboratory and Dr. Bradford Kline of National Security Agency.

The workshop will be held on September 21, 2023 (10am – 3pm) in conjunction with the National Cybersecurity Education Colloquium in the Chicago area.

Submission of Application

All interested CAE-R faculty members are invited to submit an application. Applicants are limited to one per CAE-R institution. For details on the application process, please contact your institution’s CAE-R Point of Contact. All applications will be reviewed and decided by the three Workshop Leads. The submission deadline is extended to July 30, 2023, at 11.59pm PT.

Travel Support

The Workshop is held in conjunction with the National Cybersecurity Education Colloquium. Travel support is available to US citizens who are selected to participate in the Workshop.

Notification of Participants

We plan to notify all participants by August 21, 2023.