China AI security requirements target developers launching products

Getting your Trinity Audio player ready...

China has announced security measures to guide the operations of artificial intelligence (AI) developers looking to launch products for commercial use.

The guidelines, published by the country’s National Information Security Standardization Committee, contain rules regulating the use of data by developers in training their AI models. Under the new rules, AI developers will not be allowed to rely on sources consisting of “over 5% of illegal and harmful information.”

Blacklisted data sources include those containing information in support of terrorism, “undermining national unity and social stability,” and running contrary to socialist ideals, according to the rules. Chinese authorities extend the blacklisting to existing data censored on the Chinese internet.

The rules place a burden on AI developers to obtain data in ethical ways. Firms are expected to seek the express consent of copyright holders before using their content to train their models, extending the same consent requirement on using biometric data in model training.

AI developers in China will need to submit a detailed security assessment to regulatory authorities before the public launch of their AI models. Regulators are expected to make their independent audits to ensure that AI firms meet the stipulated security requirement before a commercial launch.

The new guidelines follow mounting concerns trailing the rise of AI over the potential for misuse by bad actors to adversely disrupt existing processes. Aware of the risks, Chinese authorities preemptively geo-restricted ChatGPT for failing to meet national standards, resulting in the arrest of a citizen who allegedly used the tool to spread fake news.

Chinese regulators scrambled interim measures back in August to ensure safe technology usage. Although guaranteeing the protection of consumers, the rule book was designed to promote the development of home-grown AI platforms over Western platforms, with over 70 models launched within a month of its coming into effect.

The rules have also been construed as an attempt by the government to flex its regulatory muscles over the sectors akin to its handling of the digital currency sector.

AI cold war brews

The decision of U.S. authorities to place trade restrictions on the supply of AI chips to China has sparked a cold war between both countries. China hit back at the use by imposing restrictions on the export of gallium and germanium, raw materials used to create semiconductors.

China has moved to build its own AI chip factory to power its ambitions to become an industry heavyweight by 2030. The first-of-its-kind factory will utilize several lithography machines around one particle accelerator, but there remain several technical challenges standing in the way of the ambitious projects.

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.

Watch: AI & blockchain will be extremely important—here’s why