In the summer of 2026, I will be working with the Centre for the Governance of AI (GovAI) in London as a Summer Fellow on its AI governance research track.
Why AI governance? Over the past decade, much of my work as an international journalist focused on geopolitics, national security, and technology. I have reported from inside the TSMC — the world’s most advanced semiconductor manufacturer — interviewed the president of Taiwan, and embedded aboard a US nuclear-powered submarine. Much of this reporting examined the intensifying geopolitical competition between the United States and China. Over time, it became increasingly clear to me that artificial intelligence would become one of the defining issues shaping world order and US-China rivalry in the decades ahead.
Frontier AI has now become a mainstream political and economic issue. Governments and companies are racing to develop increasingly powerful systems that could exceed human intelligence and fundamentally reshape industries, military capabilities, and labour markets. Yet, much of the focus remains on accelerating AI capabilities, rather than building sufficient safety guardrails and governance frameworks to ensure these systems can be deployed responsibly and safely.
While pursuing my Master’s degree in climate policy at the University of Cambridge, I observed a rapid expansion of the AI governance field. There are growing resources to support AI safety research, alongside a recognition that the field needs talent from a wide range of backgrounds to help shape discussions. During the Michaelmas term, I completed the AI Alignment Fellowship at the Cambridge AI Safety Hub.
The stakes are enormous. Some experts consider artificial intelligence to be the most important technological shift since the industrial revolution. Others are concerned about the prospect of large-scale labour displacement. Meanwhile, there are genuine concerns about the deployment of AI systems in high-stakes scenarios, such as military decision-making and critical infrastructure. As AI systems become more efficient and widely adopted, institutions may increasingly delegate decisions to them, even when fundamental issues around training transparency and accountability remain unresolved.
I am looking forward to contributing research that can help advance discussions in this field, and I will share more updates in this space over time.
Eric Cheung 