Abstract

Cyber or biological attacks involving artificial intelligence could trigger public backlash against AI. However, people may already factor risks into their views, think regulatory costs outweigh the benefits, or believe regulation would be futile. To investigate how citizens would respond, we conducted experiments with 6,000 Americans. Learning about an attack significantly increased support for stringent regulations and reduced trust and willingness to use AI. Effects were larger for unintentional than intentional attacks, and repeated events than one-off ones. Backlash occurred even when attacks could have happened as easily without AI. Surprisingly, punishment did not depend on the breadth and depth of harm. Our findings suggest AI attacks could slow AI by prompting stiff regulations and/or depressing consumer demand.

Biography

Michael Tomz is the William Bennett Munro Professor in Political Science at Stanford University; Senior Fellow at the Institute for Economic Policy Research and the Center on Global Poverty and Development at Stanford; and the Landreth Family University Fellow in Undergraduate Education. He holds a Ph.D. from Harvard; a master’s degree from Oxford, where he was a Marshall Scholar; and an undergraduate degree from Georgetown. He served as Chair of the Department of Political Science at Stanford from 2020-2026.

Tomz received the International Studies Association’s Karl Deutsch Award, given to a scholar who, within 10 years of earning a Ph.D., has made the most significant contribution to the study of international relations. He also won the Giovanni Sartori Award for qualitative methods, the Jack Walker Award in Political Organizations and Parties, and the Okidata Best Research Software Award, among others. His research has been supported by the National Science Foundation.