Artificial Intelligence (AI) has made remarkable progress in recent years, offering numerous benefits for society, the economy, and scientific advancement. Central to these developments is the rise of open-source AI—technologies whose source code and data are freely accessible for others to use, study, modify, and distribute. Open-source AI plays a crucial role in accelerating innovation through collaborative development, reducing redundancy, and democratizing access to AI capabilities. This accessibility fuels economic growth and social advancement, enabling individuals and organizations to customize and enhance sophisticated AI models for specific purposes. Moreover, global collaboration among researchers, developers, and users on open-source AI fosters collective progress on shared projects and promotes the establishment of guidelines and best practices for transparency, accountability, and ethics. By making code and data accessible, open-source AI enhances transparency and accountability, helping to identify and address biases, errors, and ethical concerns, and enabling users to understand how the technology operates.
Prominent nations like China and the United Kingdom have positioned themselves as leaders in the open-source AI community. For instance, the UK-based company Stability AI has developed several popular open-source generative AI tools used for creating images, audio, 3D models, and code. Similarly, China has produced some of the top-performing open-source large language models (LLMs) globally, such as Qwen (by Alibaba) and Yi (by 01.AI). These open-source AI initiatives provide substantial competition to proprietary AI, where developers restrict public access to the underlying technology.
However, open-source AI also presents unique challenges. Unlike proprietary AI, where developers can control how users utilize their technology, open-source AI, once publicly available, can be freely accessed and modified, potentially leading to misuse. Malicious actors may tamper with open-source AI to remove safeguards, manipulate results, and generate inaccurate information. Additionally, these technologies can be exploited for dangerous and illicit activities such as cyberattacks, spreading disinformation, committing fraud, creating contraband, and other illegal endeavors. Another significant challenge is the potential lack of responsible oversight in open-source AI projects. This absence of supervision can result in known bugs or security vulnerabilities remaining unaddressed, and open-source AI might be offered without warranties or guarantees, leaving users uncertain about the quality of the data used to train the models. Furthermore, the collaborative nature of open-source development can introduce risks, such as the possibility of attackers surreptitiously embedding malicious code or data into a project.
Addressing AI risks is a global concern, and both the United Kingdom and China have taken proactive steps, even as they support their respective firms’ AI development and use. The UK convened an AI Safety Summit in 2023, which saw participation from numerous countries, including China. The summit concluded with the Bletchley Declaration, where participating nations resolved to sustain an inclusive global dialogue that engages existing international fora and other relevant initiatives, contributing openly to broader international discussions. President Xi Jinping later echoed this call for mutually beneficial cooperation on common interests, including AI, during a bilateral meeting with the U.S. president in San Francisco. In October 2023, the Chinese Ministry of Foreign Affairs issued a statement advocating for global collaboration to foster the sound development of AI, share AI knowledge, and make AI technologies publicly available under open-source terms.
Despite these high-level declarations, it remains uncertain whether the United Kingdom and China can transform their aspirations for closer AI cooperation into meaningful action. To gauge the feasibility of such partnerships, it’s essential to understand whether the concerns and priorities of AI experts outside government align and their experiences to date on collaboration.
In the rapidly evolving landscape of AI, the United Kingdom and China have emerged as pivotal players in both development and policymaking. The Bletchley Declaration, signed on November 1, 2023, by 28 governments including the UK, US, EU, Australia, and China, marks a significant step in recognizing the potential catastrophic risks posed by AI. The declaration was a highlight of the AI Safety Summit hosted by the British government, where countries agreed to collaborate on AI safety research. UK Prime Minister Rishi Sunak emphasized the transformative potential of AI and the responsibility to ensure its safe development, acknowledging the serious harm advanced AI systems could cause, whether deliberate or unintentional.
Michelle Donelan, the UK technology secretary, underscored the need for collective action in addressing AI risks. Frontier AI, referring to the most cutting-edge systems, poses unique challenges as these systems could potentially surpass human intelligence in various tasks. Elon Musk, speaking at the summit, warned about the difficulty in controlling such advanced AI systems. The summit’s communique represented a diplomatic success for the UK and Sunak, who initiated the summit out of concern for rapid AI advancements without adequate oversight.
The summit also showcased a rare display of global unity, with the US Commerce Secretary, Gina Raimondo, and the Chinese Vice-Minister of Science and Technology, Wu Zhaohui, sharing the stage. Wu emphasized China’s commitment to mutual respect, equality, and mutual benefits in AI development. The declaration welcomed international efforts to promote inclusive economic growth, sustainable development, and innovation through AI, while protecting human rights and fostering public trust.
Despite these high-level agreements, significant challenges remain in achieving meaningful international collaboration on AI regulation. The Biden administration recently issued an executive order requiring US AI companies to share their safety test results with the government before releasing AI models. Vice President Kamala Harris emphasized the importance of regulating both existing AI models and more advanced ones in the future. Meanwhile, the UK and the US announced separate initiatives for AI safety research, with the US establishing an AI Safety Institute within the National Institute of Standards and Technology.
The EU is also working on passing an AI bill aimed at developing principles for regulation and specific rules for technologies like live facial recognition. However, there is still little international consensus on what a global set of AI regulations should entail or who should draft them. British officials had hoped for a more unified approach, including the possibility of using the UK’s AI taskforce to test new models globally before public release. Instead, the summit highlighted the different approaches and priorities of participating countries.
The UK summit’s success in bringing together diverse nations to discuss AI risks and collaboration is a positive step, but the road ahead is complex. The next summit, to be hosted by South Korea in six months, and another by France in a year, will be crucial in continuing the dialogue and working towards a more coordinated global approach to AI regulation.
While the United Kingdom and China have demonstrated a willingness to collaborate on AI safety, substantial challenges persist. The Bletchley Declaration and the AI Safety Summit have laid the groundwork for future discussions, but achieving meaningful international cooperation will require ongoing effort and compromise. As AI technology continues to evolve, addressing its risks through global collaboration remains paramount.
