BRUSSELS — On a warm overcast afternoon in late September, Brussels’ digerati streamed into a cramped event space, just moments from the headquarters of the European Commission, to listen to the U.K.’s man of the hour. 

Blonde and natty in a crisp white shirt and slim-fit navy suit, Matt Clifford — the British Prime Minister’s official representative for this week’s AI Safety Summit — ambled to the lectern with the smiling ease of someone who has delivered dozens of impromptu speeches. 

The event, invitation-only and held under the Chatham House Rule, was just one leg of Clifford’s globetrotting, which has taken him from London to Washington and Beijing. These days, as he told POLITICO, he “can sleep anytime, anywhere.”

Clifford has been weaving across the planet to talk to top policymakers and tech barons about this week’s Bletchley Park summit, which will focus on severe risks like AI-aided cyberattacks and weapon design, and on which Rishi Sunak has pinned his hopes for a legacy. Many tech CEOs have known Clifford for years; presidents and prime ministers had better get up to speed.

A venture capitalist, chairman of the U.K.’s moonshot factory Advanced Research and Invention Agency (ARIA), and now an AI diplomat, 37-year-old Clifford has become one of the most influential people in British tech — just as post-Brexit U.K. scrambles to become a global beacon of AI rulemaking.

The politician’s techie

Clifford’s rise neatly maps onto the parabola of the U.K.’s tech industry: from curio, to jewel in the crown, to geopolitical tool. His debut came in 2011, just as then Prime Minister David Cameron was hitching his wagon to London’s burgeoning startup scene – dubbed the Silicon Roundabout. 

A McKinsey consultant with degrees from Cambridge (medieval history) and MIT (computational statistics) Clifford yearned for a change, and a colleague handed him a report McKinsey had just published on the Roundabout recommending investment in nurturing tech founders. 

Clifford jumped at the opportunity. He had grown up in Bradford — a northern English city scarred by deindustrialization — and taught himself to code because, he said, he “didn’t want to work in [fast food chain] Gregg’s.” 

Together with fellow consultant Alice Bentinck, he founded Entrepreneur First (EF), an accelerator that invests in graduates to help them launch startups. EF would go on to build some of the U.K.’s most successful tech unicorns. 

It also gave the duo an in to attend the monthly breakfasts Cameron held in No. 10 with London’s tech grandees. 

Clifford’s affability has helped him develop a network spanning from European startuppers to Silicon Valley heavy-hitters — LinkedIn co-founder Reid Hoffman, an ex-board member of OpenAI, sits on EF’s board and prefaced Bentinck’s and Clifford’s 2022 book “How to Be a Founder”. 

“Matt is a Swiss Army knife type,” said Dom Hallas, head of British lobbying group Startup Coalition. “But he’s also just, like, a really nice guy.”

Bentnick, his EF co-founder, said that Clifford thinks up murder mystery games for colleagues to solve, and that he uses ChatGPT to write “storybooks” for his kids. 

During the pandemic, as the British tech industry teetered on the brink, Clifford worked with Hallas and others to convince the Treasury to launch an emergency £250 million startup fund. “Whether it’s regulation, incentives, the crisis moments of the pandemic or the collapse of Silicon Valley Bank, Matt has been critical for facing those challenges,” Hallas said.

Clifford became the politician’s techie and the techie’s policy wonk. “He has cachet. He is very valued in the British tech community — which is in a way also why he’s valued by political people,” said Benedict Macon-Cooney, a chief strategist at the Tony Blair Institute for Global Change. But he is still a techie at heart. Clifford has taken a sabbatical from EF and plans to return after his summit work is wrapped up.

Building a British DARPA

After Boris Johnson triumphed in the U.K.’s 2019 general election, with tech-savvy enforcer Dominic Cummings in tow, Clifford started devoting more and more issues of his weekly newsletter, Thoughts in Between, to the subject of funding advanced science research. 

He also launched a reading club focused on initiatives such as the Manhattan Project and the 1969 Apollo 11 moon landing that managed to “achieve exceptional collective output.” That was hardly by chance: Cummings (whose blog was included in the reading group’s syllabus) had made no mystery of his grand plan to create a “British DARPA” devoted to funding ambitious science projects, and it looked like he would get his way. 

When the Advanced Research and Invention Agency (ARIA) was finally announced in 2021, Clifford would have had an easy case to make in his application for the chairmanship, to which he was appointed in July 2022: not only he had invested in technology companies for a decade, but he had written extensively about how exactly the research agency should work. [Full disclosure: Clifford also wrote about ARIA in a WIRED op-ed that I commissioned as an editor back in 2020]. 

“Most of my policy work came out of that newsletter,” Clifford said. “It had three main topics: geopolitics of technology, AI, and science funding and accelerating – all my ARIA conversations originally came out of writing, week in week out, about it.” 

Writing about AI in the newsletter, which was well read among both techies and policymakers, might also have bolstered Clifford’s credentials for his current unpaid work on the summit. Likely, so did the fact he is on first-name terms with many Silicon Valley technologists building advanced AI systems. In late summer 2022, some six months before OpenAI launched its most powerful model, GPT-4, Clifford was offered an early demo that left him “mind blown.” (He declined to say exactly how he got the demo).

Clifford is enthusiastic about AI’s advantages, from better medicine to more efficient public services. But to reap those, he thinks, you first need to get the people on board — hence the summit. 

“AI is not very popular with the public,” he said. “Therefore talking about safety is not to scare the public: it’s actually to reassure them so that we can capture the benefit.” 

The summit’s own focus on tail risks, rather than present concerns such as AI-fuelled bias and disinformation, has sparked speculation that its agenda is inspired by effective altruism, a strand of utilitarianism popular in elite universities and Silicon Valley, some of whose adherents worry about evil, almighty AIs’ potential to kill off humankind.

Clifford does not count himself as an effective altruist, although he seems generally sympathetic to their cause, going as far as speaking at a global effective altruism conference in June. “I have a lot of respect for a lot of [effective altruists and their] work but I’ve always been too much of a virtue ethicist to go all-in,” he said. Indeed, during his talk at the effective altruism event, he recommended that attendees read “After Virtue” by Alasdair MacIntyre — a thinker whose worldview is hardly utilitarian.

He pushes back on the idea that the summit has been captured by the “doomer narrative” espoused by some effective altruists. “Talking of killer robots — I don’t think that’s helpful at all,” he said. “[The summit] is much more about how we avoid a misuse that turns the public so much against AI that you get a chilling effect on adoption?”

Not a ‘political animal’

The call from No. 10 asking Clifford to help with the summit came at the end of a long stretch of AI-related work. In late 2022, he helped conduct a government review of emerging technologies where the U.K. could have a crack at setting standards: Clifford put special emphasis on AI, which seems to have influenced Sunak’s thinking. 

A few weeks later, in March 2023, he was appointed to help build the U.K.’s task force focused on advanced AI, or frontier models, and in May he orchestrated the meeting between Rishi Sunak and the CEOs of AI labs OpenAI, DeepMind and Anthropic, all of which are now on the summit’s invitation list. 

Despite once being an “ardent remainer” and Sunak being a Brexiteer, Clifford and the PM enjoy a good rapport, which Hallas said first became apparent when the two were on stage together at Treasury Connect, a conference then-Chancellor Sunak organized in 2021.

Politics rarely seems to factor into Clifford’s actions. “I’m not really a political animal,” he said. “My entire career I’ve been thinking about how to use technology as a source of leverage to make the world better.” 

But over the past few years, and especially over the past few weeks, he has learned how to talk to politicians, and to win them over. “Politicians value that — being a successful entrepreneur, being a successful investor — I know what it takes to make technology work for people,” he said. “My starting point is: how do we get things done?”