Ambassador Zheng Zeguang has praised openly released AI models from Chinese labs such as Qwen, DeepSeek and Kimi, arguing that openness and cooperation are vital. But that claim — that openness is a defining feature of China’s AI ecosystem — warrants scrutiny, and the wider lesson is that no country’s system is as open as its diplomacy suggests.
AI openness and GeoGPT’s limits
GeoGPT, a geoscience system from Zhejiang Lab showcased at last month’s World AI Conference as a model of jointly governed open science, is presented externally as open. In practice, however, it would not meet the UN-recommended model openness framework: it releases model weights (built mainly on Alibaba Qwen, whose licences are not Open Systems Interconnection-compliant), provides no training data or application source code, and its governance committee answers to Zhejiang Lab itself. That distinction matters: a free hosted service is not the same as an open one. GeoGPT users route their data through China’s legal jurisdiction, creating dependency and sovereignty risks.
By contrast, domain rivals publish more. The European Space Agency’s Earth Virtual Expert (EVE) releases its training and evaluation tooling, while the US non-profit Ai2’s OLMo publishes everything needed for reproducibility. These examples illustrate different levels of openness and reproducibility in practice.
None of the concerns outlined above are unique to China. The US private-enterprise model raises similar issues, underscoring the ambassador’s broader point: ecosystems are often less open in practice than diplomatic language implies. If the UK is to benefit from the cooperation proposed, it should press for shared openness standards — for example the UN-recommended model openness framework — as the basis for joint work. Honest measurement of openness, not just rhetoric, is where real partnership begins.
Britain’s choice: a distinctive, accountable AI
Britain cannot afford to be a passive consumer of AI while other nations shape the technologies that will define the future, but the debate should not be reduced to a geopolitical race. The key question is what kind of AI the UK wants to build. Beyond research universities and the life-sciences sector, the country’s strengths include the NHS, public institutions and a long tradition of ethical inquiry — assets that create an opportunity to develop AI that is trustworthy, transparent and directed toward the public good rather than solely commercial advantage.
Healthcare shows both promise and risk: AI can reduce administrative burdens, improve diagnosis and widen access to care, but public confidence depends on strong governance, meaningful human oversight and assurance that patient data serves citizens rather than corporate or geopolitical interests. Recent work by the Medicines and Healthcare Products Regulatory Agency’s national commission into the regulation of AI in healthcare has highlighted these concerns.
Rather than imitating Silicon Valley’s market-driven model or China’s state-led approach, Britain could offer a distinctive path combining innovation with democratic accountability, scientific excellence and social justice — a form of leadership aligned with national traditions and one that others might choose to follow.
