Tensions rise between China and the West on the regulation of artificial intelligence before the global AI security summit hosted this week by British Prime Minister Rishi Sunak. Several prominent Chinese AI experts joined their Western counterparts in signing a bold declaration warning of the “existential risk” posed by advanced AI and calling for strict international oversight.
According to an article in the Financial Times on November 1 report, the declaration, signed by figures including Andrew Yao of China and Yoshua Bengio of Canada, calls for the creation of a regulatory body to mandate the registration and auditing of advanced AI systems. It also advocates “shutdown” procedures and requires developers to devote 30% of their budget to security. These proposals go further than the regulations currently proposed by the US, EU and UK.
China is ready to support strict regulation of AI, even if it focuses on AI for censorship purposes.
China’s willingness to support aggressive AI regulation contrasts with its domestic push to use AI for censorship and social control. This indicates that Beijing could take a tough stance on AI safety at the summit, competing with Washington’s priorities of banning AI-related discrimination and protecting privacy. This rivalry comes amid broader tensions over global technological leadership.
However, the draft communiqué from the British summit, consulted by the Financial Times, does not propose specific regulations. It simply warns of the potential catastrophic harms of AI. Sunak wants the non-binding document to be signed by the United States, India, South Korea, Japan and others at the summit. Participants include technology leaders such as Sam Altman of OpenAI and Elon Musk of Twitter.
In a bid to cement the summit’s legacy, the UK has reportedly announced that South Korea will host the next AI security summit in 2024. Analysts view South Korea as a neutral choice amid the AI regulatory tensions between China and the West.
The competing visions of AI oversight at Bletchley Park highlight the difficult task of forging a global consensus on AI risk management. With advanced AI on the horizon, the divide between East and West on AI safety regulations continues to widen.
Featured image credit: Aleksandar Pasaric; Pixels; THANKS!