Agentic AI and Cybersecurity: Threats, Governance, and Strategy

5th Floor
1557 Massachusetts Ave.
Cambridge
Fri., Apr. 3, 2026, 12:20 – 1:20 p.m.



The same capabilities that make AI valuable for cybersecurity, including autonomous operation, rapid decision-making at scale, and minimal human oversight, are also the ones most likely to cause harm. These risks are further complicated by gaps in existing legal and policy frameworks, making regulation and accountability particularly challenging.

This panel discussion will drill down on several questions at the heart of this paradox: what does agentic AI mean for the threat landscape and offense/defense balance in cybersecurity; how should defenders think about the opportunities and drawbacks of employing agentic systems; and who bears liability when AI causes or fails to prevent a breach?

This is a hybrid event. Please be sure to RSVP for in-person attendance (Harvard ID holders only), or register to join on Zoom. Lunch will be served for in-person attendees.

Gazette Classification: Community, Information Technology
Organization/Sponsor: Berkman Klein Center for Internet & Society at Harvard University | Sponsored by Orrick, Herrington, & Sutcliffe LLP through the Orrick Colloquium on CyberSecurity/CyberLaw at Harvard Law School.
Speaker(s): James Mickens, Associate Professor at Harvard University, Josephine Wolff, Professor at the Fletcher School and Tufts University, Robert Knake, Venture Partner at Paladin Capital Group, Fred Heiding, Post-Doc at the Belfer Center’s Defense, Emerging Technology, and Strategy (DETS)
Ticket Web Link: cyber.harvard.edu…
Contact Info: [email protected]
Harvard Key Required: No
More info: cyber.harvard.edu…