Five days. That is how long Matt Clifford held both jobs before the arithmetic caught up with him.
On September 2, 2026, Anthropic named Clifford its Managing Director for International Affairs, a role leading the company’s work with governments outside North America. He reports to Tino Cuéllar, Anthropic’s Chief Global Affairs Officer, joining a policy and global affairs team of roughly 70 people. On September 7, 2026, Clifford announced he was standing down as chair of the Advanced Research and Invention Agency (ARIA), the UK R&D funding agency he has chaired since 2022.
“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.
The resignation is the least interesting part of this story. The interesting part is that the arrangement it undoes had already been examined and cleared. The Department for Business, Innovation, Science and Trade formally reviewed Clifford holding both posts and signed off on a set of “robust mitigations” — ethical walls, recusal protocols, the standard governance toolkit. That package did not survive a week of parliamentary attention.
The structural problem is easy to state and hard to paper over:
- ARIA holds a commercial licence with Anthropic, alongside other AI providers. The agency is a customer.
- Clifford now works for the supplier, in a job whose entire function is dealing with governments.
- ARIA spends public money on frontier research, and AI touches nearly every programme it funds.
Dame Chi Onwurah, chair of the House of Commons Science, Innovation and Technology Committee, put it bluntly when the dual role first surfaced, calling it a “clear conflict of interest” and adding: “It’s hard to see how it’s possible to mitigate potential conflicts of interests, given that AI is of almost universal relevance to advanced research and innovation.”
Clifford is not a peripheral figure in British technology policy. He wrote the UK’s AI Opportunities Action Plan and previously served as the Prime Minister’s AI adviser — which is precisely why the appointment drew scrutiny that a less prominent name might have escaped. His defence of the original arrangement was that he had “agreed to do this, with appropriate safeguards against potential conflicts in place.” The safeguards existed. They simply were not persuasive.
He will remain ARIA chair until November 6, 2026, staying on at the government’s request while the agency searches for a successor — an interim arrangement agreed with the department.
Onwurah welcomed the outcome while declining to close the file. “It’s right that the conflict of interest between the taxpayer funded ARIA and Anthropic has been addressed through Matt Clifford’s decision to step down as ARIA’s Chair,” she said, “however, important questions remain.”
Those questions are the real story. Not why Clifford left, but how a government department looked at a public research agency chaired by an employee of one of its AI vendors and concluded that a memo about ethical walls would do the job. The revolving door between AI labs and the governments meant to regulate them is not new. What is new is how little friction it now meets on the way through — and how quickly public pressure, rather than process, turned out to be the only working brake.
