Michael Veale: Regulating small AI models demands global attention now

Michael Veale, a professor of technology law and policy at University College London, speaking during the Panel Session "New Intellectual Conditions and the Re-foundation of Humanities and Social Sciences" at the 2025 Beijing Forum, emphasized a critical and often overlooked need to focus on the misuse of small AI models. He argued that significant risks arise from models deployed on devices or via cloud services, which are accessible enough for ordinary individuals to run on standard computers and can be weaponized to cause real-world harm, enabling manipulation and fraud.

In his interview at the 2025 Beijing Forum, co-sponsored by Peking University, the Beijing Municipal Education Commission, and the Chey Institute for Advanced Studies, Professor Veale further contended that these risks are not sufficiently considered in international policy discussions. These fora tend to focus predominantly on the grand challenges posed by large frontier models. While acknowledging the importance of these advanced systems, he warned that they should not consume all the regulatory attention, leaving a dangerous gap in addressing the immediate threats posed by widely accessible, smaller-scale AI.