AI swaps faces to commit fraud, AI covers star songs, and AI "resurrects" the deceased...... With the continuous development of AI technology, it is no longer difficult to "clone" other people's faces and voices online, and related disputes are not uncommon.
Not long ago, the Beijing Internet Court held a first-instance judgment in the country's first case of infringement of personality rights in AI-generated voices, clearly determining that the scope of protection of natural persons' voice rights and interests can be extended to AI-generated voices on the premise of being recognizable. The determination of the recognizability of AI-generated voices should comprehensively consider the use of the actor, and whether the ordinary audience in the relevant field can recognize it as the criterion for judgment.
Experts interviewed by reporters believe that the research and development and application of AI technology need to be carried out on the basis of protecting individual rights and interests. The safety and reliability of AI technology should be improved to prevent sound from being illegally obtained and used; It is also necessary to strengthen industry self-discipline, establish a sound regulatory mechanism, regulate the use of AI technology, and ensure that AI technology develops in accordance with the concept of people-oriented and technology-oriented.
Speech synthesis is real
Impersonating an acquaintance to commit fraud
The reporter searched in a variety of AI software and found that there were many cases of imitating the voices of well-known artists. Last year, "AI Yanzi Sun" covered songs such as "Hair Like Snow", which quickly became popular on the Internet, with more than one million views on multiple platforms, and the popularity remained high. After that, a group of other "AI singers" appeared, including the AI of deceased singers. Subsequently, a large number of AI translation works emerged, such as Guo Degang speaking cross talk in English.
It is understood that the core technology used by "AI Yanzi Sun" comes from the Sovits4.0 model. It is a free learning software based on AI technology, users only need to enter lyrics and tunes, and they can accurately simulate human voices and automatically generate a cover song.
The reporter searched on a short video platform and found that there are many learning tutorials about this kind of technology. Some industry experts pointed out that at present, through AI technology, it is possible to tamper with, forge and automatically generate images, sounds, and videos, resulting in a false effect, and some criminals are likely to use this type of technology to carry out illegal and criminal acts, endangering the public interest.
Among them, the most common method is to impersonate acquaintances to commit telecom fraud. Not long ago, Ms. Yang from Jurong, Jiangsu Province, transferred 35,000 yuan to the scammer's account after receiving multiple calls from her "daughter" asking for the registration fee. In this regard, the police handling the case repeatedly reminded: "You must be cautious when encountering transfers, and seeing is not necessarily believing." ”
"The voice of a natural person is also a type of personality right and is protected by the Civil Code. No one shall deface, deface or falsify the voice of others, let alone imitate the voice of another person without his or her permission to make a profit. Xu Hao, a lawyer at Beijing Jingshi Law Firm, said.
Xu Hao introduced that unlike copyright, the voice of a natural person is unique, and it is highly recognizable in multiple dimensions such as timbre, intonation and pronunciation style.
"The voice of a particular person can produce the attribute of a certain economic value, and being violated will not only cause personality damage, but also economic loss." Lian Dayou, director of Beijing Jingben Law Firm, said that in practice, such cases usually face the dilemma of difficulty in protecting rights, and there are problems such as difficulty in obtaining evidence, presenting evidence, and making identification.
The consent of the right holder has not been obtained
Unauthorized use constitutes infringement
The "resurrection" of the deceased by AI is also one of the most controversial topics.
The reporter saw on an e-commerce platform that many online stores specialize in the business of AI "resurrecting" the deceased, saying that they can "let dynamic photos speak, repair and synthesize a smile to customize a memorial video", and the general price is 10 yuan. However, it is written on the product details page that please communicate with customer service before shooting, and the specific price will be determined according to the degree of difficulty.
According to the customer service, it only costs a few dozen yuan to simply make the photo move, and it costs hundreds of yuan for a slightly more complicated expression. If you need to use AI face swapping technology, the price ranges from a few hundred yuan to tens of thousands of yuan depending on the similarity, and the similarity of its voice and appearance can reach more than 95%.
It is understood that AI "resurrection" of the deceased is based on advanced technologies such as big data analysis, speech synthesis, and machine learning, and through deep learning of the text, audio, and video materials of the deceased before his death, he creates a virtual image that is very similar to his voice and smile.
According to Zhu Wei, deputy director of the Communication Law Research Center of China University of Political Science and Law, whether the legal authorization of the right holder is obtained is the key to determining whether there is infringement of voice rights and interests. In recent years, the application of AI speech synthesis technology has become more and more common, but editing and using the voice of a natural person requires the separate consent of the person concerned. If the person concerned has passed away, the authorization of the relevant rights holder should be obtained, but this may also raise ethical issues.
Zhu Wei believes that the so-called "resurrection" is essentially the use of deep synthesis technology to generate so-called virtual humans or digital humans, which need to use various data such as the face and voice of the deceased to produce corresponding content for their own purposes. If the perpetrator "resurrects" his relatives for the purpose of mourning, in principle, it does not constitute infringement, but if it is for illegal purposes such as profit-making, attracting attention, or fraud, it may constitute infringement or even a criminal offense.
"If fans and fans use AI technology to resurrect idols out of love for deceased idols, this behavior is suspected of violating the law." Lian Dayou said that according to Article 994 of the Civil Code, if the name, likeness, reputation, honor, privacy, and body of the deceased are infringed, his spouse, children, and parents have the right to request that the perpetrator bear civil liability in accordance with the law; Where the deceased has no spouse or children and the parents are deceased, other close relatives have the right to request that the perpetrator bear civil liability in accordance with law.
Lian Dayou said that the act of using technology to "resurrect" the dead may also involve intellectual property issues, such as "resurrecting" deceased idols, often using related intellectual property rights such as songs, films, television, and works. The relevant intellectual property rights may belong to other companies, businesses, etc., and they also need to obtain their authorization before they can be used.
Clarify the liability for infringement
Formulation of technical standards and specifications
As for the infringement risk behind AI speech synthesis technology, relevant departments and online platforms have begun to regulate and standardize, trying to strike a balance between technological development and legal protection.
In November 2022, the Cyberspace Administration of China and other departments issued the Provisions on the Administration of Deep Synthesis of Internet Information Services, requiring that no organization or individual use deep synthesis services to produce, reproduce, publish, or disseminate information prohibited by laws and administrative regulations; Where it might cause confusion or misidentification to the public, it shall be conspicuously marked in a reasonable location or area of the information content generated or edited.
In July 2023, the Cyberspace Administration of China (CAC) and other authorities issued the Interim Measures for the Administration of Generative AI Services, which marks the establishment of specific administrative regulations for generative AI.
Article 7 of the Measures stipulates that generative AI service providers shall carry out training data processing activities such as pre-training and optimized training in accordance with the law, and require that data and basic models with legal sources must be used; where intellectual property rights are involved, the intellectual property rights enjoyed by others in accordance with law must not be infringed; where personal information is involved, the individual's consent shall be obtained or other circumstances provided for by laws and administrative regulations shall be complied with; Effective measures should be taken to improve the quality of training data and enhance the authenticity, accuracy, objectivity, and diversity of training data.
In September 2023, the Ministry of Science and Technology and 10 other departments issued the Measures for the Ethical Review of Science and Technology (for Trial Implementation), which lists a list of scientific and technological activities that need to be reviewed for ethical review, including the research and development of algorithm models, applications and systems with social mobilization capabilities and social awareness guidance capabilities.
At the same time, relevant platforms are also trying to strengthen governance. In May 2023, Douyin released a platform specification and industry initiative on AI-generated content. For creators, anchors, users, merchants, advertisers and other platform ecological participants to apply generative AI technology on Douyin, the platform will strictly punish the illegal use of AI technology once discovered.
Lian Dayou believes that there is still room for further refinement of the rule of law in the field of AI. It can accelerate the provision of AI laws and promote the establishment of generative AI norms at the national level. For difficult legal issues that arise in existing cases, it is necessary to gradually form a consensus through academic discussion and judicial practice, and summarize and establish rules for the application of law.
"In the context of the continuous development of AI technology, it is necessary to pay more attention to protecting personal privacy and voice rights, and at the same time, it is also necessary to standardize the application scope and use of AI speech synthesis technology." Xu Hao believes that for relevant companies and individuals, they should strengthen their understanding and application of AI technology, follow laws, regulations and industry standards, and avoid infringing on the rights and interests of others.
Zhu Wei called on technology suppliers to pay attention to technical ethics and formulate unified technical standards and norms to ensure that technology application has a bottom line. In terms of legal supervision, we will improve the precision, transparency and stability of prevention and control measures.
Lian Dayou suggested that in judicial practice, it is necessary to further refine the relevant legal provisions, clarify the scope of the right to sound, the criteria for determining infringement, and the liability and way of bearing. At the same time, it is also necessary to strengthen publicity and education, raise the public's awareness and attention to the right to sound, and create a good atmosphere for the whole society to jointly safeguard the right to sound.
Source: China Popular Law
![Pain! It hurts more than youth pain literature! Doctors remind: this disease is contagious, prevent → as soon as possible[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)