ACM FAccT 2025 Keynote Speakers

Regulating AI: Lessons from Emerging Legal Frameworks

Nathalie Smuha, KU Leuven Faculty of Law and Criminology

Regulators around the world have started turning their gaze to artificial intelligence, exploring whether and how the law can address its individual, collective and societal implications. While some jurisdictions are relying on existing legal tools, others, like the European Union, have opted for new AI-specific legislation. But what does it really mean to regulate AI? What challenges and tradeoffs arise when regulators seek to govern a technology that evolves faster than the regulatory process itself? And how can legislation help ensure that AI is developed and deployed in line with values like human rights, democracy and the rule of law? To answer these questions, this keynote explores the early lessons that can be drawn from emerging legal frameworks on AI, with particular focus on the newly enacted European AI Act as a case study.

Prof. Dr. Nathalie A. Smuha is a legal scholar and philosopher at the KU Leuven Faculty of Law and Criminology, where she examines legal and ethical questions around digital technologies and their impact on human rights, democracy and the rule of law. She is the author of Algorithmic Rule by Law: How Algorithmic Regulation in the Public Sector Erodes the Rule of Law and the Cambridge Handbook on Ethics and Policy of Artificial Intelligence (both with Cambridge University Press, 2025). She has taken up Adjunct Professorships at NYU School of Law and Columbia Law School, and held visiting positions at Oxford University, the University of Chicago and the University of Birmingham. Prior to her academic turn, she worked at the European Commission, where she coordinated the High-Level Expert Group on AI international organizations like the Council of Europe, UNESCO and the OECD on AI policy. She holds BA and MA degrees in both law and philosophy from KU Leuven, a PhD in law from KU Leuven, and an LLM from the University of Chicago.

Are we winning yet? FAccT, AI governance, and the shape of what comes next.

Suresh Venkatasubramanian, Brown University

We created FAccT as a space for rigorous critique, deep interdisciplinarity, and structural honesty about the role of technology in society. In the years since, the field — and the world around it — has changed in ways both thrilling and troubling. AI governance, once a speculative thought experiment, is now a high-stakes arena marked by policy experiments, institutional improvisation, and the colliding rhetoric of hype and harm. And the work at FAccT and other venues constitutes a living breathing paradigm for analyzing, critiquing, and governing sociotechnical systems.

That paradigm now faces new pressures. We find ourselves in a governance moment where large language models destabilize assumptions faster than institutions can adapt, and where the demand for structural critique must coexist with the need for operational pathways. What does meaningful governance look like in a world remade by generative AI? What does it mean to succeed — or fail — in shaping it? And how do we, as a community, continue to build something that is both honest and useful?

Suresh Venkatasubramanian is a Professor of Computer Science, Data Science, and the Humanities at Brown University, where he directs the Center for Technological Responsibility, Reimagination, and Redesign (CNTR). A pioneer in algorithmic fairness, his award-winning work has shaped the field of responsible AI, influencing how we understand and govern automated decision-making systems. Suresh has played a leading role in AI policy, serving as Assistant Director for Science and Justice in the White House Office of Science and Technology Policy, where he co-authored the landmark Blueprint for an AI Bill of Rights. He sits on the boards of the Data & Society Research Institute and the Partnership on AI, and his public scholarship has been widely recognized — including being named to Fast Company's AI20 list. His insights have been featured globally, with appearances on NPR's Science Friday, and he has contributed to policy efforts in cities and states across the U.S., as well as with the ACLU.

Making AI Evaluations That Matter

Kristian Lum, Google DeepMind

As artificial intelligence becomes deeply integrated into our social, professional, and personal lives, the methods we use to evaluate these technologies have failed to keep pace. In this presentation, I examine the current landscape of AI evaluation (particularly as it relates to concepts like bias and fairness) and argue that the dominant paradigms of AI evaluation that drive model development fail to capture the most serious real world impacts they are designed to assess. I propose several measures to build more realistic model evaluations and argue that by better integrating research on human and societal impacts into benchmark-style evaluations, we have an opportunity to meaningfully guide models towards behaviors that promote (or at least don't prevent) positive human and societal outcomes.

Techno-optimism, human pessimism, and the worlds we imagine together

Molly Crockett, Princeton University

Techno-optimists claim that superintelligent AI systems will soon surpass all human capabilities. In this talk, I will argue that the utopian visions of techno-optimists depend on pessimism about humans. "If race has no meaning to the origins of cognitive science and show how these standard methods for measuring human cognition exclude a broad range of human capabilities while playing to the strengths of AI systems. When these methods are used as benchmarks, they fake perceptions of AI capabilities while narrowing our view of human possibilities, ultimately constraining our collective imagination in ways that constitute novel forms of epistemic injustice. As an antidote, I make a case for investing in a humanistic cognitive science with more inclusive and expansive concepts of human capabilities.

Molly Crockett is a Professor at Princeton University in the Department of Psychology & the University Center for Human Values. Her research explores the psychology of moral decision-making, including social media and AI. Crockett's work integrates theory and methods from psychology, neuroscience, philosophy, and data science. Her interdisciplinary research studies, and has been published in leading scientific journals including Nature, Science, Nature Human Behaviour, and Psychological Science. Crockett co-directs the Future Values Initiative, an interdisciplinary program that supports scholarship in the intersection of ethics and science and technology and builds community among scientists and humanists at Princeton. In 2025-26 Crockett will be a member of the School of Social Science at the Institute for Advanced Study.