Prerequisite: The seminar is by admission, and open to students with established expertise in frontier AI issues or exceptional analytical capability with commitment to rapid specialization. It will meet throughout the year on an irregular schedule (dates TBD), and beyond framing classes with experts in the three areas of legal focus, will operate in a decentralized manner, with students working primarily with other students within their area of legal focus. If you would like to apply, send a CV and statement of interest to asmith@law.harvard.edu by Friday, August 29th.

Exam Type: No Exam

As AI systems begin to exceed human capabilities, they enable broader and potentially unprecedented national security risks—from AI-enabled bioattacks; to cyberattacks on power grids, financial networks, and communication infrastructure; to loss of control of advanced AI systems. The aim of this seminar will be to assess whether the U.S. government has adequate legal authority to prepare for, prevent, respond to, mitigate, or recover from these and other potentially grave scenarios, and if not, to identify legislative reforms that would correct any weaknesses so identified.

The seminar will bring law students together with legal and policy experts to develop the legal critique. The class will be divided into three areas of legal focus–cybersecurity, lawful governmental surveillance, and preparedness and response authorities, especially under the Defense Production Act. Experts with significant government experience will introduce seminar participants to each of these legal fields in a standard classroom setting and then students will dive deeper under a professor’s supervision. Students will also participate in tabletop exercises and workshops facilitated by RAND’s Technology and Security Policy Center to stress-test existing and proposed legal mechanisms in scenarios including AI loss of control and AI-enabled cyber incidents affecting critical infrastructure. Based on the initial classes, student research, and the tabletop experiences, students within each area will develop a rich understanding of existing statutory capabilities, and identify gaps in legal authority that weaken the opportunity of the executive to address AI risks. The objective in each section will be to craft a paper that identifies strengths and weaknesses within that domain, and proposes legislative reform as needed.

The credit breakdown for this seminar is as follows: two total credits with one credit awarded in the fall and one credit awarded in the spring.