British Prime Minister Rishi Sunak played host to a significant gathering of world leaders, tech CEOs, and researchers during the first-ever global Artificial Intelligence (AI) safety summit. The two-day event witnessed influential figures like Elon Musk and OpenAI’s Sam Altman mingling with dignitaries such as U.S. Vice President Kamala Harris and European Commission chief Ursula von der Leyen to delve into the future of AI regulation.
Leaders from 28 nations, including China, signed the Bletchley Declaration, a joint statement acknowledging the risks associated with AI. Both the United States and the United Kingdom unveiled their plans to establish AI safety institutes, and two more summits were scheduled to occur in South Korea and France in the coming year.
Despite some consensus on the necessity of AI regulation, disagreements persist regarding the specifics of how to implement it and who should spearhead these efforts. Policymakers have increasingly focused on the risks posed by rapidly evolving AI technologies, particularly since the release of ChatGPT by Microsoft-backed OpenAI last year. This chatbot’s remarkable ability to engage with human-like fluency has prompted calls for a potential pause in the development of such systems, amid concerns that they could gain autonomy and pose threats to humanity.
While Prime Minister Sunak expressed excitement over hosting Tesla-founder Elon Musk, European lawmakers cautioned against excessive concentration of technology and data in the hands of a few U.S.-based companies. French Minister of the Economy and Finance Bruno Le Maire emphasized the risks associated with a single country dominating the AI landscape, stating that it would be detrimental to all.
A notable departure in approach was the UK’s proposal for a more relaxed AI regulation strategy, diverging from the EU’s AI Act, which is nearing finalization and seeks to impose stricter controls on developers of “high risk” applications. Vera Jourova, Vice President of the European Commission, attended the summit with the aim of promoting the EU’s AI Act. She stressed the need for global rules, even if other countries don’t adopt the EU’s laws wholesale, in order to ensure that the democratic world has a role in shaping AI governance.
Despite projecting an image of unity, attendees acknowledged that the three dominant power blocs at the summit—the U.S., the EU, and China—sought to assert their dominance. Some suggested that Vice President Kamala Harris upstaged Prime Minister Sunak by announcing the U.S. government’s AI safety institute shortly after Britain had made a similar announcement. Harris also focused on the technology’s short-term risks, contrasting with the summit’s primary focus on existential threats.
China’s presence at the summit, as well as its endorsement of the “Bletchley Declaration,” was hailed as a positive development by British officials. China’s vice minister of science and technology expressed the country’s willingness to collaborate on AI governance, emphasizing that all nations, regardless of their size, have equal rights to develop and use AI.
However, tension between China and Western nations surfaced when Wu Zhaohui, the Chinese minister, stated that countries, regardless of their size, have equal rights in AI development. Wu participated in the ministerial roundtable but did not attend the public events on the second day of the summit.
During behind-closed-door discussions, the potential risks of open-source AI emerged as a recurring theme. Some experts raised concerns that open-source AI models could be exploited by malicious actors to create chemical weapons or super-intelligent systems beyond human control. Elon Musk, speaking at an event in London with Prime Minister Sunak, emphasized the challenge of addressing open-source AI that approaches or exceeds human-level intelligence, leaving questions about how to manage it.
Yoshua Bengio, an AI pioneer tasked with leading a “state of the science” report commissioned as part of the Bletchley Declaration, underscored the high-priority nature of addressing the risks associated with open-source AI. He stressed the need for protective guardrails to safeguard the public while dealing with the open-source release of powerful AI systems.