Matt Clifford resigns as ARIA chair over Anthropic conflict of interest

On September 2, 2026, Anthropic announced that Matt Clifford - the outgoing UK AI policy architect who had chaired the Advanced Research and Invention Agency (ARIA) since its 2022 founding - would join the company full-time as Managing Director for International Affairs, running government relations across Europe and Asia-Pacific ^[1]. Clifford's original plan was to hold both jobs at once: keep the ARIA chair while drawing a paycheck from a frontier AI lab whose interests ARIA's taxpayer-funded research grants could directly touch - a plan the departure reverses ^[2]. That arrangement lasted five days. Dame Chi Onwurah, who chairs the House of Commons Science, Innovation and Technology Committee, initially called the dual-role arrangement a 'clear conflict of interest' ^[3]. By September 7, Clifford had reversed course, announcing his resignation as ARIA chair on LinkedIn with a line that reads as much like damage control as principle: 'I have decided to step down to ensure my new role at Anthropic doesn't become a distraction from ARIA's incredible work.' ^[2]He isn't leaving immediately - at the request of business and science secretary Jonathan Reynolds, he'll stay on in an interim capacity until November 6, 2026, to give the department time to launch a search for a permanent successor ^[2], with what he described only as 'appropriate safeguards against potential conflicts in place.' ^[4]Clifford's own account of the move stayed diplomatic, framing it as protecting ARIA from a 'distraction.' At least one independent commentator read the same sequence more sharply, describing it as Clifford being forced out once MPs flagged the conflict of interest - a gap between the official framing and how outside observers read the timing.

Clifford's episode isn't happening in isolation - it's the latest, and most visible, entry in a pattern critics have tracked for months: senior UK government figures who shaped AI policy moving directly into AI company payrolls, including former prime minister Rishi Sunak (to Anthropic and Microsoft), former chancellor George Osborne (who joined OpenAI in December 2025) ^[5], and former deputy prime minister Nick Clegg (previously to Meta) ^[3]. Tom Brake, CEO of the advocacy group Unlock Democracy, argues the pattern amounts to a transfer of sensitive policy knowledge into private hands: 'Transferring sensitive policy knowledge to private firms could undermine the public interest.' ^[3]What makes Clifford different is that his arrangement got publicly unwound before it could fully take effect - a resignation forced by parliamentary scrutiny rather than a quiet transition. Whether that counts as the system working, or as proof it only responds after the fact, is exactly the question Onwurah's committee is still pressing.

Even with the resignation announced, Onwurah isn't finished asking questions. 'It's right that the conflict of interest between the taxpayer funded ARIA and Anthropic has been addressed,' she said ^[2]- but 'important questions remain' about how the dual-role arrangement was approved in the first place ^[6]. Her committee wants a fuller account of what conflict-of-interest assessments were undertaken before Clifford's Anthropic appointment was announced, and what safeguards actually protect ARIA's governance going forward ^[3]. In the meantime, Clifford stays on as chair while simultaneously drawing an Anthropic salary and overseeing an agency that directs hundreds of millions of pounds toward frontier research ^[7]. ARIA chief executive Kathleen Fisher, who took the top operational role in February 2026, runs day-to-day operations while the chair search plays out ^[4].

The reason this resignation drew parliamentary attention - rather than passing as a routine personnel change - is the scale of influence Clifford had accumulated. He became ARIA's founding chair when the agency launched in 2022 ^[6], then went on to lead the Frontier AI Taskforce and the UK's 2023 AI Safety Summit ^[2]. In January 2025, he authored the Department for Science, Innovation and Technology's AI Opportunities Action Plan for the Starmer government ^[1]. By the time he announced his Anthropic move, ARIA itself had active research and development funding agreements totaling £514.1 million, per its 2025-2026 annual report ^[7]. That combination - architect of national AI strategy, chair of a taxpayer-funded frontier-research agency, and now a paid international affairs executive at one of the labs whose commercial fortunes that strategy helps shape - is what turned a personnel announcement into a conflict-of-interest story.