69% of people in the United Kingdom are worried about the economic impact of AI job losses according to a 2026 survey from King's College London. These worries need not materialise. The AI boom is projected to generate phenomenal levels of economic value, and there is currently tremendous uncertainty surrounding how this windfall, estimated to reach £550 billion by 2035, will be redistributed towards generating benefits for a broad section of society.
Since the First Industrial Revolution, governments have relied on taxing labour for their revenues. Rand estimates that 66% of US federal revenue in 2024 came from directly taxing labour income. As AI erodes income tax revenues, we situate token taxes within the existing landscape of policy options for taxing AI.
Token taxes are one mechanism of AI value capture. Rather than taxing labour, this policy focuses on the point of value generation in AI: where the models process inputs and outputs – tokens – that drive AI usage. Since the price paid per token is approximately similar to wages paid for human labour, a token tax can be designed to approximate an AI income tax.
Can compute governance infrastructure be leveraged to reliably audit token taxes?
What are the legal challenges associated with implementing token taxes?
What are the advantages and disadvantages of token taxes compared to alternative taxation mechanisms such as compute taxes, VAT, and digital services taxes?
Can we model the impact of a token tax on the UK economy using LLM-powered agent-based modelling?
A UK MP has written expressing interest in a token taxes memo, with further letters of interest from parliamentarians expected.
Read the letter (PDF) ↗Token taxation on AI inference appears as a policy plank in the Steyer gubernatorial manifesto.
Read the plan ↗The token tax position paper was cited in an interview by US Congressman Greg Casar.
Watch the video ↗The people behind the project

Lucas Irwin
Author of Position: Token Taxes Can Mitigate AI's Economic Risks (ICML 2026). He is currently a DPhil student at the University of Oxford and the Oxford Martin School AI Governance Initiative (AIGI). He holds a bachelor's in Computer Science from Princeton University, was a summer fellow at the Centre for AI Governance, and has had his work featured in Bloomberg and Just Security.
Project Lead

Simone Gargiulo
Conducts research on technical AI governance, with a focus on whether physical and hardware-based measurements can help verify AI compute and distinguish different types of LLM activity. He is currently a Senior Fellow at Pivotal Research, and previously worked at CERN and EPFL, where he completed his PhD in physics. He later contributed to the ESA Galileo programme, working on atomic-clock technology used in Europe’s global navigation satellite system.
Technical Governance Researcher

Shahil Goodka
A technology lawyer from London, most recently an Associate in RPC LLP’s leading commercial and technology practice. He was previously a secondee in Meta’s legal team, advising on global cybersecurity incidents and data governance. Shahil now collaborates on policy research as a SPAR Research Fellow and advises AI safety non-profits.
Legal Researcher

Saskia Poulter
A DPhil student at Nuffield College, Oxford, where she works on the political economy of housing. Aside from her degree, she is a policy analyst at several UK think tanks and consultancies, focussing primarily on growth policy. She holds the ESRC’s Advanced Quantitative Methods studentship.
Economics Researcher

Akansh Jain
Has built his career building systems from scratch across big tech in e-commerce, marketplaces, and finance. He brings an implementer’s view to the token taxes project, which sits at the intersection of his work on climate tech, Global South alignment, and AI safety through an Apart Research initiated project.
Technical Researcher

Professor Philip H.S Torr, FRS, FREng
Professor Torr is a British scientist and a professor at the University of Oxford. His research interests are in machine learning and computer vision and he is a fellow of the Royal Society, fellow of the Royal Academy of Engineering, fellow of St Catherine's College, Oxford, and distinguished research fellow at the Institute for Ethics in AI at Oxford.
Advisor

Sharif Kazemi
Sharif is a MATS Research Fellow working on AI assurance and verification. Previously, he was Policy and Strategy Lead at the World Bank’s applied AI lab, advising governments on operationalising national AI strategies. He was a Fulbright Scholar at Columbia University's School of International and Public Affairs.
Advisor

Andrei Potlogea
A Lecturer in Economics at the University of Edinburgh. Originally a trade economist, he now works on the economics of AI. He contributed to Epoch AI’s GATE project, an early integrated assessment model of the AI transition, and his current research examines the economic impact of AI regulation.
Advisor
Whether you are interested in funding us or working on research related to taxing AI, we would like to hear from you.
thetokentaxproject@outlook.com