Five days
Matt Clifford is standing down as chair of ARIA, the UK’s taxpayer-funded high-risk research agency, five days after Anthropic announced him as its managing director of international affairs. He announced the decision on 7 September and will remain in post until 6 November to give the government time to find a successor.
“Having completed my first full term last month, I have decided to step down to ensure my new role at Anthropic doesn’t become a distraction from ARIA’s incredible work,” he said.
What he was doing on both sides
Clifford is the closest thing the UK has to an author of its AI policy. He ran the 2023 AI Safety Summit and wrote the AI Opportunities Action Plan, commissioned in July 2024 and published in January 2025; the government adopted all 50 recommendations and appointed him its AI Opportunities Adviser.
ARIA is modelled on the US DARPA and hands public money to high-risk, high-reward research, a large share of it AI-related or AI-enabled. Anthropic’s new role puts him in charge of the company’s dealings with governments outside North America.
The original plan was to keep both
When the appointment was announced on 2 September, Clifford intended to stay on as ARIA chair, with a stated plan to recuse himself from Anthropic-related matters at the agency. He also retains a role at the venture firm Entrepreneurs First.
Dame Chi Onwurah, who chairs the House of Commons Science, Innovation and Technology Committee, said his intention to remain at ARIA “creates a clear conflict of interest.” That is the argument that won.
What the committee still wants
Onwurah welcomed the resignation but did not treat it as the end of the matter, saying important questions remain about what conflict of interest assessments were carried out and what safeguards exist to protect confidence in ARIA’s governance.
The question is not really about Clifford’s intentions. It is about a recusal-based arrangement being judged sufficient at the point of appointment, by whoever judged it, for a person who chairs the body that decides which UK research gets public money and simultaneously represents a frontier lab to governments. The committee wants to know who signed that off.
The wider pattern
Governments have spent three years hiring people who understand frontier AI, and the people who understand frontier AI mostly come from, or go to, the labs. Britain built its AI institutions faster than most countries partly by drawing on that revolving door. This is the bill arriving: the same expertise that makes someone worth appointing is what makes the appointment hard to defend once they take a job on the other side.
What to watch
Who replaces Clifford at ARIA, and whether the government publishes the conflict of interest assessment the committee is asking for. Also worth watching is whether the next appointment comes from outside the AI industry entirely — which would be a change of practice, not just of personnel.
