Verity Harding Warns AI 'Arms Race' Framing Risks 2-Superpower Lock-In
Updated
Updated · WIRED · Jul 8
Verity Harding Warns AI 'Arms Race' Framing Risks 2-Superpower Lock-In
1 articles · Updated · WIRED · Jul 8
Summary
Harding argues in a new essay anthology that treating AI as an arms race is steering policy toward rivalry, making cooperation on safety and shared benefits harder.
2016-2020 experience briefing leaders at DeepMind underpins her case that AI policy once leaned more international, before ChatGPT's 2022 debut, the pandemic and the Ukraine war hardened US-China competition.
Trump administration moves — nationalist AI rhetoric and export-control pressure that led Anthropic to pull a frontier model — are cited as signs the worst-case dynamic is already emerging.
Middle powers such as Canada, France, Japan, South Korea, India and the UK could form a coalition, Harding says, to avoid being forced into a binary US-China alignment.
Her broader warning is that race language concentrates power in governments and top labs, leaving smaller states as 'vassals' and producing less safe AI systems.
With AI already transforming modern warfare, is abandoning a competitive stance a dangerous risk to national security?
Could shared catastrophic AI risks force the US and China into an unprecedented alliance for mutual survival?
Can a coalition of middle powers forge a third path, or must they ultimately align with an AI superpower?
Moving Past the AI Arms Race: Lessons from History and Policy for Cooperative Global AI Development
Overview
In 2026, the global conversation around artificial intelligence is dominated by the 'AI arms race' narrative, which drives countries like the United States to enforce strict export controls and prioritize technological sovereignty. This competitive mindset has led to nationalistic AI development, with policies aimed at maintaining an edge over rivals such as China. However, this approach is increasingly challenged by concerns about the societal impact of rapid AI expansion, especially among younger generations. The report highlights that focusing solely on competition risks missing opportunities for collaboration and inclusive governance, which are essential for managing AI’s profound effects on society and ensuring its benefits are widely shared.