Court warns of legal danger in AI-generated content
The generation-z-dominated Chinese video platform Bilibili has sprung to life recently with artificial intelligence (AI) singers, but the intellectual property (IP) team at the Minhang District court said that there are many legal issues involved in such content.
On the platform, someone synthesized the voice of former US President Donald Trump and recorded it singing classic patriotic songs, such as "I Love You, China." Other netizens faked the voice of crosstalk artist Guo Degang rapping.
After the "dry sounds" of a person without instruments, accompaniment, reverberation or harmony are input into AI software, a similar voice can be generated to say anything including singing. Many such songs with the names and photos of celebrities have earned hundreds of thousands of hits online with some videos even reaching one million views.
Although no lawsuit has yet been filed, according to laws and regulations such as the the Civil Code and Copyright Law, such unauthorized videos are likely to infringe on the celebrities' rights to name and voice, the performers' rights and the copyright of songwriters, Ni Jingjing, a member of the court's IP team told Shanghai Daily.
Many videos are labeled with disclaimers, stating that they are for entertainment only and if there is any infringement, they are willing to delete them. But Ni noted that the determination of copyright infringement is unrelated to profitability.
Ni also said the platform is likely to bear legal responsibility according to comprehensive factors such as the platform type and monitoring and supervision capabilities.
In addition to AI singers, many virtual characters and related works became popular in recent years, such as AI celebrities and models.
For example, Liu Yexi, an AI celebrity on the short video platform Douyin, is a "virtual beauty talent who can hunt monsters in the metaverse," according to the character's self-introduction.
The character and its Chinese fantasy short play, produced by Shanghai-headquartered technology company Create One, has attracted more than 8.35 million followers since 2021.
Liang Zikang, chief executive of the company, said in a previous report on thecover.cn that 10 more AI characters will be launched by the end of this year to establish the company's metaverse.
The technology may also be used to infringe upon the rights of ordinary people.
Qingpu District's procuratorate reminded the public not to be deceived by the words of the offenders, who use high-tech marketing, such as AI and metaverse, to lure others into investing. It launched a suit against a pyramid-selling case involving the metaverse on Tuesday, involving over 11.6 million yuan (US$1.65 million) and more than 14,000 victims.
A man named Liang Zhizhong and another two people developed a metaverse system SpayX in 2021 and cheated others, claiming it was a "development research think tank of China," the procuratorate found.
The three declared that the systems in the cities of Shenzhen and Beijing had been completed and they needed the public to invest in the system in Shanghai, the procuratorate said.
The investment was based on the virtual currency Tether tokens and people who invested in the project were promised they would enjoy preferential treatment plus a high profit if they attracted more members.
Last year, Liang and his group even held a raffle during a business meeting to build up their credibility. The star prize was a car and an apartment and all the winners were confederates of the men.
After the investors found that they could not contact the men, they called the police. The trio were arrested within two months.
The procuratorate has taken legal action and further investigation is underway.
To promote the benign development and application of AI technology, departments and companies have recently issued several relevant regulations.
For instance, Douyin published its platform specification about AI-generated service last week, saying that the contents made by AI should be marked clearly and publishers were responsible for the consequences of AI-generated contents. It would also impose strict penalties if users created and published AI-generated content that infringed on rights or regulations.
Shanghai passed the nation's first city-level law and regulation about AI in October and provided the AI industry with a huge development space within the legal framework. It said if the local department found minor offenses in the field of AI, the related people or company could be exempt from administrative penalties but needed to be criticized and educated.
Last Wednesday, the State Internet Information Office completed canvassing opinions for the first national regulation on AI-generated service, which is expected to be launched this year.
In a draft called "Generative Artificial Intelligence Services Management Guideline" published online for public opinions, it required generative AI should obey general human ethics; the algorithm design, covering training data selection and model generation and optimization, should prevent discrimination on race, ethnicity, faith, country, gender, age or occupation; and AI services should avoid unfairness, fake news and leakage of personal information and business secrets. Meanwhile, AI-generated pictures and videos were required to be marked as such.
China is not the only country to take such actions. Italy has temporarily banned ChatGPT and launched a probe over associated suspected breaches of privacy rules.
In the United States, on Tuesday, the developer of ChatGPT OpenAI cheif executive Sam Altman asked members of a Senate Judiciary subcommittee hearing to regulate AI.
"My worst fears are that we cause significant – we, the field, the technology industry – cause significant harm to the world," he said in a report by China Daily. "We think that regulatory intervention by governments will be critical to mitigate the risks of increasingly powerful models."
The AI Act, the world's first legislation governing AI, will be issued soon after members of the European Parliament voted last week to advance draft regulations on rules about facial recognition, biometric surveillance and other AI applications.