Owen Cotton-Barratt

Owen Cotton-Barratt is a mathematician who researches how to act well under uncertainty about the future of AI, with a focus on existential risk and epistemics. He holds a DPhil in mathematics from Oxford, was Lead Researcher at the Global Priorities Project and Director of Research at the Centre for Effective Altruism, and modelled AI safety trade-offs at the Future of Humanity Institute. There he founded the Research Scholars Programme, whose first cohort started in 2018. In 2021 he co-authored "Truthful AI" with Owain Evans and others, and in 2023 he sat on the board that spun Elicit off from Ought. He is a research consultant at Forethought, where he co-wrote "AI Tools for Existential Security" in 2025 and writes on AI and epistemics.

Now at: Forethought
Works: 14
Citations: 203
h-index: 8
Badges: Bridge

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2014

Leading the Global Priorities Project

He holds a DPhil in mathematics from the University of Oxford and has worked as a Research Fellow at the University of Southampton. In 2014, he is Lead Researcher at the Global Priorities Project, a new effort to prioritise between broad areas for policy makers and philanthropists.
By 2016

Modelling AI safety trade-offs

By 2016, he is Director of Research at the Centre for Effective Altruism in Oxford, listed with the Future of Humanity Institute as well. His project "Decision-relevant uncertainty in AI safety" models the trade-offs of AI safety work aimed at different time horizons, and the Future of Life Institute lists $119,670 as the amount recommended for it.
2018

AI safety priorities at FHI

In 2018, he works with colleagues at the Future of Humanity Institute and the Centre for Effective Altruism on cause prioritisation for AI safety, modelling trade-offs between different types of research. As a Research Fellow at FHI, he also argues in a new paper that researchers on dangerous projects should take out liability insurance.
2018–2021

Founding the Research Scholars Programme

He founds the Research Scholars Programme at the Future of Humanity Institute, a salaried two-year programme that gives early-career researchers time to work out which questions to tackle; its first cohort starts in 2018. In December 2020, he discusses epistemic systems and layers of defence against global catastrophes on the 80,000 Hours Podcast. In 2021, he directs the programme as its third cohort starts.
October 2021

Co-authoring Truthful AI

In October 2021, he is one of eight authors, led by Owain Evans, of "Truthful AI: Developing and governing AI that does not lie". The paper proposes a standard of avoiding "negligent falsehoods", institutions to evaluate AI systems before and after deployment, and training AI systems to be truthful.
September 2023

Ought spins off Elicit

In September 2023, he sits with Owain Evans and Paul Christiano on the independent board of Ought that sells its IP to Elicit, a public benefit corporation spun off from the nonprofit. Elicit has raised $9 million by then, and Ought expects minimal operations for the next few years.
March 2025

AI tools for existential security

With Lizka Vaintrob, he publishes "AI Tools for Existential Security" with Forethought in March 2025. The paper argues that epistemic, coordination-enabling and risk-targeted AI applications can help navigate existential risks and can be accelerated.
As of October 2026

Research consultant at Forethought

He is a research consultant at Forethought with over a decade of experience working on the future of AI. In March 2026, he is one of six authors of "The Artificial Self: Characterising the landscape of AI identity", a paper on how identity boundaries for AI systems that can be copied or edited shape their incentives and risks. In April 2026, he joins Rose Hadshar on Forethought's ForeCast podcast to discuss whether language models are good enough to power tools that improve public discourse.