A billboard for AI (artificial intelligence) is seen at the World Conference on Artificial Intelligence (WAIC) held in Shanghai, China on July 6, 2023. Ally Song—Reuters
China on Thursday announced new regulations for generative AI, the technology behind OpenAI’s ChatGPT and Google’s Bard chatbot. The regulation governs all publicly available chatbots and is overseen by China’s top internet regulator, the China Cyberspace Administration (CAC). Generative AI research and technology developed for use in other countries are exempt from regulation.
Chinese tech giants such as Alibaba and Baidu have yet to release generative AI tools to the public. Experts believe they were waiting for the government to announce the final regulations. (Thursday’s policy is titled ‘Interim Measures’, but is open to change). A Chinese version of the generative AI chatbot and image generator is still in development or being trialled by B2B customers, according to a CNN report. For example, Alibaba last week released a text-to-image generator called Tongyi Wanxiang, which is still only available for beta testing for enterprise customers. And Chinese search engine giant Baidu released Ernie Chatbot to only about 650 enterprise cloud customers in March.
Developers are also required to register their algorithms with the Chinese government and undergo a “safety evaluation” if their services are deemed to have “social mobilization capabilities” that can influence public opinion. . At least initially, the policy appears to be adhered to. China’s existing censorship activities for online conversations.
The new law features an overarching requirement to “adhere to the core socialist values”. The same section of the regulation endlessly outlines the illicit use of generative AI. Some were aimed at protecting citizens, while others were aimed at increasing government control over nascent technology, such as promoting terrorism and banning the dissemination of “obscene pornography.” , must not use generative AI. ““Subvert state power”, “damage national image”, “undermine national unity”.
Domestic national security concerns related to AI have echoed at the highest levels of the Chinese government as well. According to a PBS report, Chinese President Xi Jinping, at a meeting in May, called for a “new pattern of development with a new security architecture” to address the “complex and difficult situation” that AI poses to national security. ” called out.
Thursday’s rules were drafted by the CAC but were approved by seven other agencies, including the Ministry of Education, the Ministry of Public Security and the State Radio and Television Directorate, according to the CAC website. Such broad national agency involvement lends some credence to the idea that the government wants to use AI in virtually every industry in the country, which is also outlined in the new policy. It has been. The new regulations were introduced amid an AI arms race erupting between China and the United States. identified as an issue. luckreported by Nicholas Gordon.
Chinese Regulations Guide AI Regulation
Thursday’s regulations were an update to interim guidelines issued in April that tech companies found too restrictive. They are now providing the US and other countries with blueprints on how to address some high-profile issues surrounding generative AI, including potential piracy and data protection.
These include some of the world’s first explicit requirements for generative AI companies to respect intellectual property rights. The topic recently surfaced in the US when comedian Sarah Silverman sued OpenAI and Meta for using her copyrighted work to train machine learning models.
The CAC’s new policy was also required to outline specific privacy rights for individual users. If users disclose their personal information while using the service, China’s generated AI platform will be responsible for protecting their personal information. In addition, if a company plans to collect or store information that is otherwise protected, we will provide our users with terms of use to “determine the rights” they have when using the Platform. Must provide. Terms of service are widely used in technology applications ranging from social media to app stores, but are not yet required by law for U.S. generative AI platforms, according to a May congressional report. Additionally, according to the regulations announced by the CAC, all existing Chinese privacy laws will apply to AI. These provisions may be particularly helpful for the United States, which currently does not have comprehensive data protection laws.
The recently announced measures also provide clues about China’s global AI ambitions, specifically the policies that will eventually be used to regulate AI use around the world. Developers and suppliers, as well as chip makers, were “encouraged” to participate in “making international rules related to generative artificial intelligence,” according to the new law.
The idea that China wants comprehensive regulation has been debated in the past, most recently by Tesla CEO Elon Musk. He predicted on Wednesday that China would be open to a “cooperative international framework for AI regulation” and said he had spoken to officials during a recent visit to China.
