Dispatches from the evolving digital political economy
Taiwan Internet Governance Forum
Taiwan will hold its own Internet Governance Forum a week before the UN IGF in Kyoto, Japan, which Taiwanese will not be allowed to attend. Organized around the theme “Critical Moment: Fragmentation, Geopolitics, AI Revolution and Resilience,” the program has IGP Director Milton Mueller speaking on Cyber Sovereignty and moderating a panel discussion of Web PKI that includes EFF’s Aaron Gables, Internet Society’s Amreesh Phokeer, and Dr. Tsung-Min Kuo of the national telecom operator. The event also features a keynote speech by Paul Vixie on “DNS resilience and security,” a panel discussion on the status of Internet governance with William J Drake, Jeanette Hofmann, Milton Mueller, Anriette Esterhuysen, and Kenny Huang, and presentations by Kuo-Wei Wu, Chairman of the Taiwan Internet Governance Forum, Kenny Huang, Chairman of the Taiwan Internet Information Center, Audrey Tang, Minister of Digital Development, Chun Lee, Deputy Minister of Foreign Affairs, and ICANN board members Becky Burr and Edmon Chung. The event will be held at the Fubon International Conference Center in Taipei. The full agenda can be seen here.
Labor and AI: an Early Encounter
Although it was not the only precipitant of the Writers Guild of America (WGA) strike, AI was a major point of contention between the writers and studios. The union’s new contract establishes limitations on how AI can be used in the writers’ room. A summary of the agreement establishes the following baselines:
- AI can’t write or rewrite literary material, and AI-generated material will not be considered source material…, meaning that AI-generated material can’t be used to undermine a writer’s credit or separated rights.
- A writer can choose to use AI when performing writing services if the company consents and the writer follows company policies, but the company can’t require the writer to use AI software…when performing writing services.
- The Company must disclose to the writer if any materials given to the writer have been generated by AI or incorporate AI-generated material.
- The WGA reserves the right to assert that exploitation of writers’ material to train AI is prohibited by MBA or other law.
This can be seen as an assertion of writers’ property rights over the use of their literary output to train AI applications. It also shields writers from losing screen credit and payments to AI derivative work. This would provide some protection against AI as a substitute for the traditional role of writers in TV and motion picture production. But it does not preclude more radical applications of AI that might obviate the need for writers altogether.
China Clarifies and Relaxes Data Security Laws, But…
China’s Cyberspace Administration of China (CAC) released for public comment some draft Regulations for Cross-border Data Flow on September 28. The draft is an attempt to clarify and relax previous data exit regulations in China, such as the Measures for Outbound Data Transfer Security Assessment 《数据出境安全评估办法》and Measures for Standard Contracts for Cross-Border Transfer of Personal Data《个人信息出境标准合同办法》. These regulations required companies exporting certain kinds of sensitive data – called “important data” and “personal information” – to get a security assessment from the CAC. But as Kendra Schaefer pointed out, the state had not yet defined in detail what kind of data counts as “important data.”
The new Regulations introduce 8 categories of cross-border data that are exempt from the requirement of a security assessment or other restrictions on personal information. The introduction of these exemptions reflects the recognition by Chinese authorities that a “one-size-fits-all” approach to data exit is impractical. The strict control over low-sensitivity cross-border data was hindering international trade, production, and collaboration. The new CAC regulations serve as a positive signal, allowing multinational enterprises to engage in international business within China and bolstering the confidence of overseas investment in China.
According to Article 7 of the Regulations, pilot free trade zones have the authority to independently establish a “Negative List” which specifies the data subject to outbound data transfer security assessment, model contract for outbound personal information transfer, and personal data protection certification within the free trade zones. Data that is not included in the Negative List can be transferred abroad without undergoing the aforementioned assessments, contracts, or certifications. This indicates that China continues to explore data flow management through the implementation of pilot free trade zones. Moving forward, the development of cross-border data management methods within these zones will be closely observed. Since 2020, Beijing, Shanghai, Zhejiang, Hainan provinces have all introduced independent plans for Free Trade Zone regulating cross-border data flow.
However, it is important to note that the Regulation still contains certain ambiguities. Firstly, the definition of “critical data” in categories 1-2 remains unclear. Secondly, Regulation category 3 which pertains to “personal information not collected in China” still carries some ambiguity. It is uncertain whether it refers to the absence of personal information collected within China or the geographical location of the entity processing the personal information. Thirdly, there is ambiguity within categories 7-8. For instance, if an organization processes personal information for one million individuals but only provides the personal information of less than 10,000 individuals abroad within a year, is that data allowed to directly leave the country? Moreover, it is not specified whether “expected to supply overseas within one year” refers to existing data or newly generated data. Additionally, the regulation lacks explicit mention of how sensitive personal information should be handled, with only a limited reference to personal information in this context.
Overall, the regulations indicate a shift towards a more relaxed approach to regulating cross-border data flow in China. However, the actual impact and effectiveness of their enforcement remain to be seen. Currently, the regulations are still in the public comment phase, and based on previous experiences, the official document is likely to be published in November.
Japan Embraces AI to Fight Disinformation
In a previous Narrative, IGP described Chinese-state media influence operations concerning Japan’s Fukushima water release. Japan’s response to the disinformation is to utilize artificial intelligence (AI) for the purpose of monitoring, assessing, tracking, and countering foreign disinformation campaigns on social media platforms.
The Japanese Ministry of Foreign Affairs have been consistently posting videos and written explanations in English, Mandarin, and Korean to clarify why the release of treated water poses no environmental threat. Building upon these efforts, the Ministry introduced an AI-powered system in the spring of 2023 to detect false information circulating on social media. In June, the system effectively identified a false claim originating from South Korea claiming that a high-ranking Japanese Foreign Ministry official had made a substantial political donation to the International Atomic Energy Agency (IAEA). During a press conference on July 23rd, Yoshimasa Hayashi, the Minister of Foreign Affairs, highlighted the ministry’s utilization of AI for the identification of disinformation and fact-checking purposes. He also mentioned that the Ministry regularly asks social media platforms to remove the content.
Japan’s plans for combating disinformation extend beyond merely identifying false information regarding the Fukushima case, however. On August 1st, the Cabinet Office of Japan announced the government’s intention to create a more advanced AI system capable of discerning information manipulation and facilitating the evaluation of distortions in facts by relying on concrete evidence. Japan’s Economic Security Legislation also plans to allocate funding to both public and private research institutions to develop AI technologies geared towards “active cyber defense.” This includes capabilities such as situational awareness, identifying system vulnerabilities, and detecting cyberattacks. In a subsequent development, on August 14, the Japanese Ministry of Foreign Affairs officially declared its commitment to allocate 500 million Yen to the 2023 fiscal year budget to further bolster their initiatives aimed at countering foreign influence operations using AI technologies.
Because Japan’s policy discussions regarding the development of counter-disinformation AI tools are still in their early stages, it may be premature to predict the precise outcomes and ramifications of this framework. Is this leading to the proverbial “censorship industrial complex”? IGP recommends that Japan adopt a prudent and thoughtful approach when shaping its counter-disinformation AI system. It should preserve the editorial freedom of social media platforms, encompass a well-defined definition of what constitutes a national security threat, and ensure that the framework remains in alignment with the principles of freedom of expression. For instance, while specific and concrete scientific evidence may exist in cases like Fukushima, it may not always be straightforward to delineate the boundary between disinformation, misinformation, and debatably inaccurate information in the broader context. It is imperative to establish a robust definition of disinformation security threats and clearly define the specific objectives of deploying counter-disinformation AI tools.
“Media Freedom in Hong Kong” event divides UN
On September 27, the UK organized a side event at the UN’s 54th Human Rights Council on Media Freedom in Hong Kong. China’s government denounced the event as interference in its “internal affairs” and urged “all esteemed Permanent Missions to refrain from participating in this event in any way.” Nevertheless, the event gained co-sponsorship from Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Iceland, Italy, Japan, Latvia, Liechtenstein, Lithuania, Luxembourg, Netherlands, New Zealand, Norway, Slovakia, Sweden, and the United States of America.
Before the sovereignty transfer in 1997, China agreed to allow Hong Kong political autonomy for fifty years under a framework known as “one country, two systems.” Those guarantees were tossed out after the passage of a National Security Law in 2020. Since then China has directly suppressed critical journalism, jailed journalists and publishers, prevented public commemorations of Tiananmen Square, and generally intimidated any dissent. The UN General Assembly elected China as a member of the Human Rights Council in 2020.
