Generative artificial intelligence, represented by ChatGPT, is triggering a new round of artificial intelligence revolution. Generative AI has changed the way information and knowledge are produced, reshaped the way humans interact with technology, and opened up our imagination for AI in general. It is foreseeable that generative AI will have a significant impact on industries such as healthcare, education, retail, finance, consulting, media, and gaming. And like any other technological innovation in the past, generative AI is a double-edged sword, helping prosperity and development while also generating false and harmful information, invasion of privacy, algorithmic discrimination, infringement of intellectual property rights, and other risks. How to find a balance between technological innovation and public interest, and between development and security, is a difficult problem of the times posed by new technologies and applications such as generative artificial intelligence to all mankind. The Interim Measures for the Administration of Generative Artificial Intelligence Services (hereinafter referred to as the "Measures") jointly promulgated by the Cyberspace Administration of China and relevant departments is China's answer to the questions of the times. As the first special legislation for generative artificial intelligence in the world, the Measures take encouraging innovative development as the keynote, taking the legitimate rights and interests of individuals, public interests and national security as the bottom line, and coordinating development and safety in terms of principles, concepts and regulatory scope, which is not only China's exploration of regulating generative artificial intelligence, but also opens up a new path for global artificial intelligence governance.
1. Encourage innovation and standardize the principles and concepts of development
Article 1 of the Measures clearly states that "promoting the healthy development and standardized application of generative AI" as the first goal, and Article 3 further states: "The state adheres to the principle of attaching equal importance to development and security, promoting innovation and governing according to law, takes effective measures to encourage the innovative development of generative AI, and implements inclusive and prudent and categorical supervision of generative AI services", highlighting that the Measures put the innovation and development of generative AI in an important position, and is in line with the EU's Artificial Intelligence Responsibility Directive (AI). There are differences in legislation centered on risk prevention and accountability, such as the Liability Directive and the draft AI Act. In June 2023, the European Union added corresponding regulatory standards for general artificial intelligence. However, practice has long proved that artificial intelligence, a new technology with continuous development and uncertainty, needs to adopt a more scientific and inclusive and prudent governance approach.
In the era of rapid changes, non-development is the biggest insecurity, and development is the foundation and key to solving all problems on the mainland. As a key general-purpose technology, generative AI is important for people's well-being, economic development, national security, and strategic competition. However, there is still a gap between the mainland and advanced countries in terms of relevant basic research, technological ecology, industrial development, and top talents. In April 2023, the Politburo meeting of the CPC Central Committee specifically pointed out that it is necessary to attach importance to the development of general artificial intelligence, create an innovation ecology, and attach importance to risk prevention. Adhering to this spirit, the Measures require, on the one hand, in view of the characteristics of generative AI technology, the management departments to improve scientific supervision methods that are compatible with innovation and development, so as to realize the synchronous evolution and synergy of technological innovation and regulatory innovation (Article 16). For example, Article 7 fully considers the massive and heterogeneous nature of training data, and takes "authenticity, accuracy, objectivity and diversity of data" as the optimization goal, which fits the technical characteristics of large models; On the other hand, encourage industries to explore and optimize generative AI application scenarios in various industries and fields, support the development and collaborative innovation of multiple subjects such as industry organizations, enterprises, scientific research institutions, public cultural institutions, and professional institutions (Article 5), and promote the sharing and development of chips, software, frameworks, computing power, algorithms, data resources, and other full-links and ecosystems (Article 6).
2. Inclusive and prudent and categorical regulatory scope
As the concept and principle throughout the Measures, "encouraging innovation and standardized development" is not only reflected in Chapter II "Technology Development and Governance", but also in Chapter III "Service Specifications" and Chapter IV "Supervision and Inspection and Legal Liability".
Article 2 of the Measures clearly anchors the scope of application of the Measures on "providing generative AI services to the domestic public", which is concise and profound, which is the focus of this Law. In detail, this means, first of all, that the Measures distinguish between "service" and "technology". This distinction is not only connected with the "service"-based regulation of the Provisions on the Administration of Internet Information Service Algorithm Recommendation and the Provisions on the Administration of Deep Synthesis of Internet Information Services, but also leaves room for trial and error for technology research and development and application, and avoids excessive entry into the iterative field of technology itself and affecting technological innovation. From a broader perspective, the regulation that emphasizes "service" rather than "technology" is precisely conducive to jumping out of the theoretical dispute between technology neutrality and constructionism, starting from the system theory, and incorporating the evolutionary system, organizational form, and application scenarios in which artificial intelligence inhabits into the regulation, so as to achieve both scenario-based and holistic governance. Second, the Measures distinguish between "public-facing" and "non-public-facing". On the one hand, from the perspective of risk prevention, "facing the public" may lead to risk expansion and accumulation, especially the spread of false and harmful information will disrupt the order of cyberspace communication, market order and social order, so it has become the focus of supervision; On the other hand, generative artificial intelligence is not only a tool for interacting with ordinary user information, but also a powerful assistant to improve efficiency and enhance competitive advantage within enterprises. According to a survey by consulting firm Accenture, nearly six in ten companies intend to use ChatGPT for learning purposes, and more than 40% are willing to invest in it. Through customized models and data fine-tuning, generative AI will disrupt traditional working models and support industries and businesses. Therefore, leaving room for "non-public-facing services" within and vertical fields will help accelerate the large-scale implementation and wide application of generative AI technology on the enterprise side, help enterprises significantly improve production efficiency, and release innovation potential. Finally, the Measures coordinate "onshore" and "overseas", respond to the needs of overseas enterprises to provide generative AI services, require compliance operations under the premise of complying with this Law, and make bridging provisions for foreign investment in generative AI services, reflecting the policy orientation of inclusiveness and openness.
At present, a new round of scientific and technological revolution and industrial transformation led by artificial intelligence is in the ascendant. As a pioneer of general-purpose artificial intelligence, generative artificial intelligence will likely profoundly affect human economic, political, social, cultural and military fields. In the face of this global challenge, China has taken a step ahead. We look forward to the implementation of the Measures accumulating Chinese experience for subsequent AI legislation, contributing Chinese wisdom to global AI governance, and jointly building a community with a shared future in cyberspace.
Author: License, Director of the Digital Economy and Legal Innovation Research Center, University of International Business and Economics
Source: "Netinfo China" WeChat public account
Reviewed: Li Jiamin
Editor: Wang Linchen
Proofreader: Chen Jindan
Source: NCIT China
![What are the security challenges posed by new technologies such as generative AI? The forum was held in Shanghai[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)