via Wired AI
I Met With China’s Top AI Experts. They’re Freaking Out, Too
## The AI Arms Race Has Chinese Researchers on Edge
In June 2026, I sat down with some of China’s leading artificial intelligence researchers. The conversations were striking — not for their confidence, but for their deep unease. Across the board, these experts are worried about an accelerating AI arms race between China and the United States, one they fear could lead to what one researcher called a “Chernobyl moment” for AI: a catastrophic failure that triggers a global backlash.
## Shared Anxiety Across Geopolitical Lines
Chinese AI scientists, many of whom work at top universities and state-backed labs, express concerns strikingly similar to their American counterparts. They worry about the rapid deployment of increasingly capable AI systems without robust safety testing, the lack of international governance frameworks, and the potential for AI to be weaponized or to cause large-scale societal disruption.
“We are racing toward a cliff, but neither side can afford to slow down first,” one Beijing-based researcher told me. This sentiment echoes warnings from US-based AI safety advocates who have long called for a pause on frontier model training. As of 2026, no such pause has materialized — instead, both nations have accelerated development, pouring billions into next-generation models.
## The Specter of a ‘Chernobyl Moment’
The metaphor of a Chernobyl moment recurs frequently in AI risk discourse. Just as the 1986 nuclear disaster reshaped global energy policy overnight, a major AI incident — say, a misaligned autonomous system causing mass casualties or a catastrophic financial meltdown — could trigger a sudden, international regulatory crackdown.
Chinese experts are acutely aware that their nation’s command-and-control style of governance could respond to such an event in ways that close off beneficial research. “If something goes wrong, the government might simply shut everything down,” warned a researcher at a leading Shenzhen lab. “That would set us back years, and we would lose scientific talent to other countries.”
## The 2026 Context: Escalation Without Governance
In the years since the initial 2023–2024 AI boom, the landscape has shifted dramatically. By 2026, both China and the US have deployed advanced AI systems with multimodal capabilities and near-human reasoning in specialized domains. The US has maintained a lead in foundational research and chips, while China has caught up in applications, scale, and data resources.
Yet international governance remains fractured. The 2025 UN AI Summit produced only non-binding agreements, and bilateral talks between the superpowers stalled over export controls on advanced semiconductors. “We have the tools to collaborate on safety research — but the political will is missing,” said a professor at Tsinghua University.
## A Call for Collaboration Amid Competition
Remarkably, many Chinese AI experts are looking to the US for partnership on technical safety challenges: aligning AI systems with human intent, preventing long-term loss of control, and building robust evaluation benchmarks. Some have begun informal cross-Pacific exchanges, though these are circumscribed by visa policies and data privacy laws.
“The core problems of AI don’t respect borders,” one researcher emphasized. “If we don’t solve them together, we all face the same risks.”
## Conclusion: A Shared Vulnerability
My conversations left me with a sobering realization: Chinese AI researchers are not triumphalist or complacent. They are deeply aware that the technology they help build could spin out of control, and that the geopolitical rivalry makes a safe outcome less likely. The arms race is not just a race of capabilities — it is a race toward a potential disaster that neither side can prevent alone.