UK AI policy chief Matt Clifford resigns as ARIA chair after Anthropic conflict of interest row

Clifford announced last week that he was joining Anthropic, the company behind the Claude chatbot, as managing director of international affairs, leading its engagement with governments outside North America. He initially said he would retain his position as chair of ARIA, the high-risk, high-research agency modelled on the US DARPA, and step back from 'everything else'. But the move drew immediate criticism from the House of Commons Science, Innovation and Technology Committee. Dame Chi Onwurah, the committee's chair, said Clifford's intention to remain as ARIA chair while working for a leading AI company 'creates a clear conflict of interest' and called on the government to set out 'robust safeguards and clear boundaries'.

Days later, Clifford has been forced to stand down, according to reports. The Guardian confirmed that he has left his role as ARIA chair, following the conflict of interest concerns. Clifford was appointed chair of ARIA in 2022 and was commissioned by the previous government to write the 'AI Opportunities Action Plan' in July 2024, published in January 2025. He was then appointed as the government's AI Opportunities Adviser. In a post on X after joining Anthropic, Clifford said he would be 'based in London and working with governments across Europe, Asia-Pacific, and beyond', adding that 'the decisions that will shape AI's impact shouldn't be made by any one actor – it will take industry, governments and civil society working together'. The UK's AI Minister is now Kanishka Narayan, following Andy Burnham's appointment as Prime Minister.

Why This Matters

Policy integrity: The rapid departure of a senior government adviser who helped shape UK AI policy to work for a major AI company raises questions about revolving-door practices and the independence of public research funding.
Global precedent: The case could set a benchmark for how other governments manage conflicts of interest when their top AI experts move into the private sector.
ARIA's reputation: Clifford's forced resignation may affect confidence in ARIA's independence, given that it invests taxpayer money in AI-related projects.

Background

Matt Clifford was a key figure in UK technology policy, having been appointed by the government in 2022 to chair ARIA, a £800 million agency designed to fund high-risk, high-reward scientific research. In 2024, he was commissioned to write the 'AI Opportunities Action Plan', which the government adopted in full, and he subsequently served as the government's AI Opportunities Adviser. His move to Anthropic, a company that directly engages with governments on AI regulation, sparked immediate conflict of interest concerns from MPs, particularly because he intended to remain chair of ARIA while working for the firm.

Key Perspectives

MPs and the Science and Technology Committee: Dame Chi Onwurah emphasised that while movement between public and private sectors is not discouraged, Clifford's dual role as ARIA chair and Anthropic employee created a 'clear conflict of interest' that required robust safeguards. The committee's pressure ultimately led to his resignation. Matt Clifford (previous statement): Clifford argued that the decisions shaping AI's impact require collaboration between industry, governments and civil society, and that Anthropic 'takes that seriously'. He had planned to recuse himself from matters involving Anthropic at ARIA. Critics and Skeptics: Some commentators may argue that the conflict was inevitable given the close ties between AI companies and government policy, and that stronger rules are needed to prevent similar situations in the future.

What to Watch

Who will replace Clifford as ARIA chair: The government's choice will signal its approach to independence and public-private boundaries.
Anthropic's UK engagement: Clifford is now based in London and will lead Anthropic's engagement with governments; watch for any new contracts or policy influence.
Potential policy changes: The UK government may introduce stricter rules on public officials taking private sector roles in AI.

ZOTPAPER