Public AI incident database counts conflate changes in reporting propensity, deployment growth, and shifts in harm frequency per unit of exposure. These issues introduce significant uncertainties challenging public and corporate policy frameworks centred on realized risks.

We propose a simple framework that establishes clear points of inquiry, separately estimates exposure from harm-rate trends, and then classifies into meaningful trajectory categories for governance decisions.

The framework combines a structured monitoring question format (SORT) to clarify coverage decisions, a tiered estimation procedure calibrated to available evidence, and LLM-assisted incident matching against public databases.

Applied to various monitoring questions, we draw conclusions regarding the monitoring ecosystem more broadly: Providing an essential interpretative classification, determining what can and cannot be claimed, and establishing that exposure estimation is required as AI deployments become increasingly common.

Supporting experts: Peter Slattery and Simon Mylius (MIT AI Risk Initiative)

Alumni

Meet the authors

(Research Team Lead)

Isaak Mengesha

Isaak is a Postdoctoral Research Associate at the University of Oxford working on forecasting technological change.

His research examines how technological change interacts with institutional incentives and coordination constraints, shaping long-run welfare and development trajectories.

Drawing on agent-based modelling, network analysis, and empirical growth research, he focuses on decision relevant analysis for risk management, and AI governance.

Branwen Owen

Branwen is an infectious disease epidemiologist, and is now applying an epidemiological approach to AI safety monitoring. Through the AI Governance Taskforce she is developing methods to assess exposure as part of an AI incident monitoring system.

She holds a PhD in epidemiology from Imperial College, where she applied a wide range of statistical methods to examine HIV transmission among women.

She then turned her hand to transmission modelling and cost-effectiveness. Her analyses have been used to inform decisions on infectious disease control at operational, national and global levels.

Charlie Collins

A UK diplomat and civil servant of 10 years, Charlie emigrated to New York after a posting to the UK Mission to the UN.
Charlie started volunteering for an AI incident database before joining the AI Governance Taskforce, working to unlock the potential of incident tracking to help AI safety practitioners prioritise their response.
Educated as an engineer, Charlie enjoys meaningful challenges that combine technical and policy problems, and working to solve these with smart, kind people.

Tina Wong

Tina is a security engineer passionate about understanding the magic and moral questions of technology, and guiding its responsible use through policy. Her technical background and international experiences have shaped her interest in exploring the global challenges of AI governance, including in the area of AI harm classification at Arcadia Impact.

She holds a B.S. in Computer Science and a History Minor from Stony Brook University. Outside of professional work, she enjoys volunteering at non-profits on digital transformation and data science projects.

Alumni

Meet the authors

Isaak Mengesha

(Research Team Lead)

Isaak is a Postdoctoral Research Associate at the University of Oxford working on forecasting technological change.

His research examines how technological change interacts with institutional incentives and coordination constraints, shaping long-run welfare and development trajectories.

Drawing on agent-based modelling, network analysis, and empirical growth research, he focuses on decision relevant analysis for risk management, and AI governance.

Branwen Owen

She holds a PhD in epidemiology from Imperial College, where she applied a wide range of statistical methods to examine HIV transmission among women.

She then turned her hand to transmission modelling and cost-effectiveness. Her analyses have been used to inform decisions on infectious disease control at operational, national and global levels.

Charlie Collins

Tina Wong

Programme

AI Governance Taskforce

The AI Governance Taskforce is a career development programme for experienced professionals looking to transition careers into AI governance, focussed on reducing risks from advanced AI.
Participants work around existing commitments during our 12 week, remote, part-time cohorts, producing policy research in teams of 4, led by our Research Team Lead staff in partnership with recognised experts in the field. Teams write an academic-style paper and accompanying blog post to build knowledge, skills and work portfolios.

Learn more