Former DeepMind Exec Warns AI Arms Race Could Lead to Disaster

Verity Harding, who from 2016 to 2020 led global public policy at Google DeepMind and briefed political leaders such as Barack Obama and Emmanuel Macron, believes that describing the development of artificial intelligence (AI) as an arms race is not only inaccurate but also dangerous.

In her new essay anthology "Reframing the AI Arms Race," Harding, along with historian Lawrence Freedman and Japanese politician Taro Kono, argues that such rhetoric shapes policy and international relations. In her view, portraying AI as a lethal weapon risks closing the door to the international cooperation essential for ensuring the technology's safety and the equitable distribution of its benefits.

Harding points out that AI research was initially rooted in international cooperation, but over time it became shaped by rivalries—between private labs like Anthropic and OpenAI, and between the US and China. She sees the Trump administration's nationalist rhetoric and attempts at export controls as symptoms of this arms race framing.

The former DeepMind executive calls for the creation of a coalition of middle powers, which could include Canada, France, Japan, South Korea, India, and the UK. Such a coalition would give them more leverage and reduce dependence on the two superpowers.

Harding warns that if the current trajectory continues, the long-term outcome could be excessive government control and centralization of power over AI systems, along with less safe and socially beneficial technology. Smaller countries would become "vassal states," forced to choose sides between the superpowers.