news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Source: Global Times

Recently, Microsoft executive Suleiman believes that any content on the Internet that is captured in order to train artificial intelligence (AI) models to better serve humanity is fair use, and does not require the copyright owner to be paid or consented to. In his view, content on the open Internet is "free software" and anyone can copy it and use it to recreate and reproduce. This view of Microsoft executives is ostensibly about the fair use of public data, but in essence it is a challenge to data sovereignty by algorithmic power.

The development of artificial intelligence and the protection of intellectual property rights should go hand in hand. Public data on the Internet can be used to train AI, but copyright cannot be infringed. There is no doubt that the widespread application of AI is of great value to the development of human society and will push mankind into the intelligent era. We should encourage its development and give it warm applause. But at the same time, its development process cannot ignore the intellectual property protection system. Training large AI models requires feeding a large amount of data, and the high-quality data is created by humans. If the skin is gone, where will the hair be attached? The premise of artificial intelligence is "artificial". In the foreseeable future, the learning object of artificial intelligence will still be human works. Without an intellectual property protection system, the motivation and number of human creation of excellent works will also be sharply reduced, and the development of AI will inevitably be hindered. Obviously, we must balance the needs of artificial intelligence development and intellectual property protection. The importance of establishing an intellectual property protection system cannot be ignored in order to develop artificial intelligence. The essence of the views of Microsoft executives is a one-sided statement from a commercial standpoint to reduce production costs for enterprises, and a defense for the use of technological advantages to plunder the interests of others.

AI giants such as Microsoft crawl massive amounts of data for training artificial intelligence, which has far exceeded the scope of "fair use". In fact, crawling massive amounts of data for commercial purposes to train AI models is difficult to be recognized as fair use in most countries. Fair use in the US Copyright Act must first consider "the purpose and characteristics of the use of the work: whether it is commercial or for non-profit teaching purposes". This is also the reason why the New York Times sued OpenAI and Microsoft in 2023 - using copyrighted news articles to train AI chatbots without permission. Related behaviors also do not fall under fair use in China's Copyright Law. Article 24 of the Copyright Law clearly stipulates that "works can be used without the permission of the copyright owner and without paying remuneration to him". The standards for fair use are: made in special circumstances; not conflicting with the normal use of the work; and not unreasonably damaging the legitimate rights and interests of the copyright owner. The key to fair use lies in judging "special circumstances". Article 24, Item 13 of the Copyright Law stipulates that "appropriate citation", "inevitable", "small amount of copying", "not for profit purposes" and other "special circumstances" are "special circumstances". Obviously, Microsoft’s business practices cannot be considered “special circumstances”, which shows that the views of Microsoft executives are based on the company’s self-interest and ignore the business logic of laws in various countries.

my country should be wary of algorithmic power threatening data sovereignty. In recent years, the field of AI in the United States has developed rapidly and has a great technological advantage. The views of Microsoft executives also represent the current data colonialism in the field of artificial intelligence. Some AI giants use their advantages in algorithms and computing power to crawl massive amounts of global data for free to train large AI models, and then sell the trained AI products to other countries. This kind of data plunder is chauvinism in the digital age. Whether it is the Age of Navigation, the Industrial Age or after World War II, there has always been a situation in history where resources from other countries have been plundered in various ways. In the era of artificial intelligence, this form of exploitation includes free and massive acquisition of data from other countries. Faced with this new type of data infringement, my country should actively respond to prevent itself from becoming a data colony.

First of all, the best way to prevent technological exploitation is technological countermeasures. In recent years, my country's artificial intelligence industry has developed rapidly, but it still needs to accelerate the pace of technological development in all aspects and continuously improve the legal system that promotes technological development. In recent years, Europe has successively introduced the General Data Protection Regulation (GDPR) and passed the Artificial Intelligence Act to regulate AI. However, this strong regulatoryism has not brought about a major explosion in Europe's digital industry, while the leniency of US legislation and the protection policy for digital giants have helped the country's companies continue to reap global digital benefits. Therefore, the AI ​​legislative policy of developing first and then regulating is worth learning. After all, only by mastering technological advantages can my country counter the US data invasion.

Secondly, we should actively advocate data sovereignty equality and resist data chauvinism. In the era of artificial intelligence, although countries have a common need to safeguard data sovereignty, there is no unified action. my country should put forward a digital consensus with global appeal, and at the same time promote the establishment of a set of international legal norms to safeguard the data sovereignty of various countries. In terms of the implementation path, we should advocate the concept that data has a sovereign scope and algorithms involve national interests in the "circle of friends" of the "Belt and Road" mechanism and the BRICS cooperation mechanism based on the principle of friendly consultation, and establish international data arbitration courts, international digital courts and other institutions that are in line with the interests of various countries, so as to make legal preparations for possible algorithmic power disputes in the future.

Data is no small matter, and the battle over algorithms is also a competition of national strength. In the era of artificial intelligence, we should focus on the future development of the country. In terms of policy, we should actively develop artificial intelligence technology and win the computing power competition with technological advantages; in terms of legislation, we should adopt an attitude of inclusiveness and prudent supervision and introduce technology-friendly laws; internationally, we should advocate data sovereignty egalitarianism and build an international institution for data dispute resolution. Only by adhering to developmentalism can we embark on the road of national rejuvenation in the era of data hegemony. (The author is a researcher at the Zhejiang Digital Development and Governance Research Center)