A series of essays from UK experts published by the AI & Geopolitics Project (AIxGEO) at Cambridge University has warned of the risk of artificial intelligence (AI) being portrayed as an arms race.
The contributors collectively argue that reducing AI geopolitics to a race for technological supremacy risks obscuring the more complex political, economic and institutional realities that shape how the technology develops.
The experts warn that the power of this framing lies less in whether it is accurate than in its ability to become self-fulfilling, narrowing political choices and risking the very values that competition is intended to protect.
Verity Harding, director of AIxGEO, who was the first global head of policy for DeepMind, described the framing of AI as a zero-sum arms race between the US and China as “a self-fulfilling prophecy that fosters nationalism and distracts us from the potential for AI to solve shared challenges”.
She urged policymakers to draw an analogy instead of the Space Race, which she said demonstrates how political pragmatism and international diplomacy can turn a high-stakes rivalry into a rules-based framework for the benefit of all humankind.
“If world leaders believe they are in a chest-thumping, life-or-death race to an AI finish line, they will react accordingly,” she warned. “And since no one knows where the finish line is, the only rule is ‘don’t slow down, and don’t help anyone else speed up’. It transforms a tool for collective progress into a theatre of zero-sum conflict.”
Harding said there is no “finish line” when it comes to AI, and no point after which any one nation can be said to have “won”. “Even if there was some rubicon of AI dominance to be crossed, isolationism and nationalism will not make AI safer, or any nation stronger,” she said.
In her essay, Martha Lane Fox noted the importance of collaboration to achieve true success with AI. She urged AI firms and policymakers to learn from how other sectors have embedded safety and risk management as core, collective principles.
Lane Fox pointed to the aviation industry as an example of how innovation is inseparable from safety. “Every new tool or method is rigorously tested, regulated and reviewed,” she said. “Flying is statistically the safest way to travel because a culture of shared vigilance, operational discipline and institutional learning has evolved over decades.”
Unlike aviation, Lane Fox said the AI sector sees governance, safety and ethics as hurdles rather than operational essentials. “There are legitimate debates, such as those raised by the European Union’s AI Act, about how to balance safety with innovation,” she said. “Yet industries like aviation show it is possible for robust, global standards and organisational discipline to enable rather than stifle advancement.”
Viewing AI through the lens of an arms race, according to Roy Austin, director of the artificial intelligence initiative at Howard University’s school of law, risks creating a race to the bottom, which erodes civil rights, entrenches inequality and undermines democratic values. “What is needed is an AI where excluded communities become its architects, designers and stewards, and transform it into a force for collective good,” he said.
The UK government tries to talk up the benefits of AI for society, but Diane Coyle, professor of public policy at the University of Cambridge and research director at the Bennett School of Public Policy, believes what governments should be doing is reevaluating their economics to ensure they are fair for the digital age and that AI is part of the mix.
She warned that, for “middle powers”, the arms race metaphor is actively harmful as it makes sovereign interest in AI seem a matter of choosing a side. Coyle believes national economic security actually requires an understanding of the changed character of the economy in the 21st century compared with the pre-digital era.
