Our Programs
Comprehensive Risk Assessment
Public Security Policy
Geostrategic Dynamics Mechanisms
Offense/Defense Dynamics
Alignment of Dynamical Cognitive Systems
Comprehensive Risk Assessment
Existing assessment techniques fail to capture the scope of the risks we face from transformative general-purpose AI. We research and devise frameworks, methodologies, and tools to empower assessors of AI systems to scan the threat surface in a top-down manner and model the propagation of resulting risks in society.
Public Security Policy
Society needs actionable AI safety policy solutions. Through targeted advocacy, policy development, stakeholder engagement, and interdisciplinary collaboration, we bridge complex technical insights from the research community with practical risk mitigation strategies across policy ecosystems.
Geostrategic Dynamics Mechanisms
The development of a technology with unprecedented potential to alter the world is a coordination challenge which requires positive-sum responses. We use game theory, mechanism design, and complex systems modelling to analyse and propose mechanisms for multiscale international cooperation.
Offense/Defense Dynamics
AI models will be used in a variety of ways, with some uses that take advantage of people, and others that protect people from being taken advantage of. We research ways of characterizing factors that affect these dynamics, and model downstream effects, both in the general case and within salient capability areas.
Alignment of Dynamical Cognitive Systems
Powerful but poorly alignable general-purpose AI agents will need to be curtailed, controlled, or somehow made safer. Through cognitive science, complex systems theory, and dynamical systems theory, we research better ways to do this while better understanding and respecting stakeholders, context, and tradeoffs.