The AI China Report #2
Your executive summary on the geopolitics of AI & China 🇨🇳 - Issue #2
Welcome back to The AI China Report. This week we’re covering China’s recent AI regulations and how they could affect U.S. AI companies, the impact of President Biden’s diplomacy with Pacific Island nations, and a new AI tool the CIA is developing for espionage.
QUOTE OF THE WEEK
“Like our forebears during World War II, we know that a great deal of the history of the world will be written across the Pacific over the coming years”
- President Joe Biden during the White House U.S.-Pacific Islands Forum Summit (September 25, 2023)
NEWS ROUNDUP
President Biden hosted leaders of 19 Pacific Island nations last week in an effort to ensure that the Pacific Islands remain “free and open”. The White House has taken interest in establishing diplomacy in Pacific Island countries, with Biden announcing $40 million in investment for infrastructure for the islands, as China has expanded it’s military presence in the area. Beijing’s China Daily newspaper wrote about the meeting, saying that the U.S. is only taking interest in Pacific Island nations as they now see China as a formidable rival: “All of a sudden, the region has become an area of interest on their geopolitical chessboard”, the newspaper wrote. (NYT)
Nvidia, the leading U.S. chipmaker has said that the government is restricting sales of its most advanced chips to “some Middle Eastern countries” over worries that China, which is already sanctioned from purchasing Nvidia’s top-of-the-line chips, may be leveraging connections with Middle Eastern countries to acquire the advanced technology needed to improve their AI development capabilities. (DW)
The CIA and U.S. intelligence agencies are working on an internal AI chatbot similar to ChatGPT, that will allow agents to ask the chatbot questions to sift through mounds of “public data” and help find “the needles in the haystack” for global CIA spying operations. (Bloomberg)
As the United Nations General Assembly convened in New York, discussions were held about the challenges of governing artificial intelligence (AI) and the potential need for a new U.N. agency to oversee it. While the details of such an agency remain undetermined, many are concerned about international cooperation, given the rapid development of AI and the cross-border implications it might have. (Time)
The U.K., which is hosting a global summit on AI safety in November, received some backlash for inviting China, and is now considering limiting access and only inviting Chinese officials for “one or two days” as intelligence officials have increased warnings of China’s extensive spying operations in the U.K. and other Western countries. (The Guardian)
Lakshmi Raman, the CIA’s director for AI, spoke at the Politico AI & Tech Summit last week. She warned that the one of the biggest threats is China using AI to create misinformation campaigns that undermine trust in the U.S. government and spark culture wars. Notably, she mentioned that China is already doing this and government agencies are bringing on people to identify these campaigns. (Fox Business)
U.S. tech titan Intel has seen strong sales of their Guadi2 processors, which were designed specifically to comply with U.S. regulations on exports of AI chips to China. Last year, 27% of Intel’s annual revenue was generated in China. Intel has partnered with Taiwan Semiconductor Manufacturing Co (TSMC) for production of the chips.
DATA TO USE
ONE BIG STORY: CHINA’S NEW AI RULES COULD AFFECT U.S. COMPANIES
Last month the The Cyberspace Administration of China (CAC) put into effect new rules for generative AI tools, which require AI companies to register with the government and pass a security review before they can be released to the public.
The rules also require that all data used for training AI models come from “legitimate sources” and don’t infringe on national IP rights. Companies creating AI tools can also be held responsible for the content that is created on their platforms including content that advocates for terrorism, extremism, ethnic hatred, violence, obscenity, and fake or harmful information.
These rules create exposure for AI companies, as it has proven difficult to prevent generative AI tools (especially chatbots like ChatGPT) from creating content that could violate these rules.
U.S. technology companies are still ahead of China in terms of AI capability, with China’s best models being comparable to OpenAI’s GPT-3 (which is a two year old model, OpenAI has since released GPT-3.5 and GPT-4 to the public). It’s likely that these regulations could make it more difficult for Chinese companies to innovate at a speed that will allow them to catch up the the U.S.
The U.S. still doesn’t have AI regulations, but is actively working on putting together guidelines as top technology leaders like Elon Musk, Mark Zuckerberg, and Sundar Pichai of Google, among others, met in a closed door session with U.S. lawmakers earlier this month to discuss regulation.
One of the most important questions regarding U.S. regulation is how to strike a balance that allows for protection of Americans (and jobs) without hindering AI innovation to the point that the U.S. loses their lead to China or other countries.
Lawmakers will be looking closely at China and the CAC rules as they decide on the best approach for AI regulation in the U.S. The severe regulations in China could prompt the U.S. to also create stricter guidelines for its own AI companies. This would affect new AI startups and especially current tech titans like Google, Meta, Amazon, and Microsoft that are already heavily investing in and developing AI products.
Stronger U.S. regulation could lead to greater protections for Americans, but hinder companies that have been improving AI technologies at a rapid pace. However, this could introduce the risk of falling behind if the Chinese government chooses not to enforce their own regulations on certain national companies that it favors.
Know someone who should be in the know about AI and China? Share The AI China Brief 🇨🇳 with your friends and colleagues!
Have feedback on how we can improve? Want to discuss this issue’s content? We’d love to hear from you in comments below!