Former DeepMind Policy Chief Warns of AI Arms Race

Verity Harding spent years helping shape public policy at DeepMind, the British AI lab now owned by Google. In a new interview with WIRED magazine, she says plainly that the United States government's protectionist approach to artificial intelligence is a sign that the worst-case scenario for the technology's development is coming true.

According to Harding, when nations focus exclusively on outpacing their rivals, the risk grows that key safeguards and ethical considerations get overlooked. The pursuit of strategic advantage is starting to overshadow the responsible deployment of AI and the mitigation of the risks the technology carries.

The US government's nationalist approach to artificial intelligence is proof that the worst-case scenario for the technology's future is coming true - Verity Harding, former director of public policy at Google DeepMind

A Fragmented AI World

The central warning of the interview concerns the fragmentation of the global AI landscape, driven by individual countries' national interests. Harding argues that this scenario leads to AI systems being built under different ethical frameworks and safety standards from country to country.

The result is a complex and potentially unstable international environment, where the absence of unified global oversight makes it harder to ensure interoperability and control over advanced AI systems. According to Harding, the lack of shared ethical principles could deepen existing geopolitical tensions and open new fronts of conflict.

A Race Nobody Wins

The interview feeds into a broader debate over whether the rivalry between the United States and China for AI dominance resembles a classic arms race. Harding points out that the intense competition for talent, computing resources and technological advantage is itself becoming a source of danger, regardless of which side ultimately comes out ahead.

That stance differs from the tone some AI companies have struck in recent months, focused mainly on the pace of development and commercializing their models. The former DeepMind official's voice is a reminder that there is considerable disagreement within the industry itself over whether the current pace of AI deployment is safe.

A Call for Cooperation Over Competition

Harding argues for moving away from a nationalist approach toward international cooperation, shared ethical principles and robust safety mechanisms. In her view, that is the only way to avoid the harmful consequences of advanced AI development, rather than chasing a one-sided strategic advantage.

The call comes as the question of global AI governance returns to the agenda of major international organizations, including recent discussions in Geneva at the United Nations on the responsible oversight of artificial intelligence.

What It Means for Europe and Poland

For countries like Poland, which are not major players in the race for the most powerful AI models, Harding's warning carries practical weight. A fragmentation of global safety standards means companies and institutions outside the US and China could face a difficult choice between competing technology ecosystems, each governed by different ethical and regulatory rules.

The European Union, pushing its own AI Act, is trying to offer a third path built on regulation, but critics note that the pace of implementing EU rules itself lags behind the pace of the technology's development. The former DeepMind director's comments bolster the argument of those who believe that without genuine international coordination, no single jurisdiction can ensure safe AI development on its own.

Sources: WIRED (wired.com), IndiaVision (indiavision.com)