The voice of AI can be seen everywhere on online platforms today, as the saying goes, 'Before you see the person, hear their voice.'. With the continuous advancement of algorithms and the help of high-performance devices and high-precision models, the sound generated by AI has evolved from being "fake at first listen" in the past to being "difficult to distinguish between true and false" today. Many audiobooks use AI dubbing, and readers can even choose to imitate the voice of a specific person, which not only reduces costs but also adds fun. At the same time, the phenomenon of AI voice abuse also occurs from time to time. Some merchants use AI to imitate the voices of well-known people and fabricate content when promoting products on short video platforms, which not only seriously misleads consumers but also has a negative impact on the parties involved. Voice "AI cloning" is the misuse of personal voiceprint information, which serves as a biometric feature with high recognizability and can be accurately recorded by electronic means. Its uniqueness, uniqueness, and stability make it a key element in linking information with specific natural persons. According to the Personal Information Protection Law, voiceprint information is classified as sensitive personal information and is strictly protected. How can AI technology generate sounds that are "fake and fake"? The reporter learned from the interview that it relies on current deep learning algorithms to accurately extract a series of key features from the collected sound samples, including frequency, timbre, tone, speech rate, and emotional expression. Subsequently, these features are transformed into complex mathematical models and further synthesized through algorithms to generate highly realistic sounds. With only a small amount of sound material, one can quickly achieve 'AI cloning' of sound A technician who declined to be named told reporters. The reporter found through a search on the app store that a large number of software have relevant functions, and some apps have been downloaded over ten million times. The reporter chooses an app for simulation, simply read out a paragraph of text or input an audio clip, and these apps can complete the sound synthesis in just over ten seconds. As the threshold for AI to generate sound rapidly decreases, some criminals take the opportunity to misuse it for fraud and false advertising. On social media platforms, it is not uncommon to use AI technology to imitate and parody short videos of celebrities from various fields. There have been a large number of short videos imitating the voice of a well-known person through AI, some of which contain inappropriate comments and have sparked social discussions. The well-known person subsequently released a video response, stating that these events have caused him distress and discomfort, and calling on everyone to stop such behavior. In addition, some merchants use AI generated sound to mislead consumers when promoting products on short video platforms. In the first AI generated voice personality rights infringement case in China announced in 2024, the plaintiff Yin, a voice actor, accidentally discovered that his voice had been AI transformed and sold it to the public on a platform operated by a defendant intelligent technology company. Subsequently, Yin sued the intelligent technology company and other defendants, claiming that the defendant's actions seriously infringed upon his own voice rights. The court held that if the sound synthesized by artificial intelligence can be associated with the natural person by the general public or the public in related fields based on their tone, intonation, and pronunciation style, it can be recognized as identifiable. Therefore, a certain cultural media company and a certain software company used the plaintiff's voice in AI without the plaintiff's permission, which constitutes an infringement of the plaintiff's voice rights and should bear corresponding legal responsibilities. For this case, Zhao Ruigang, Vice President of the Beijing Internet Court and the second level senior judge, believed that China's Civil Code had compiled the right of personality independently, wrote the protection of "voice" into the Civil Code in the form of legislation for the first time, and explicitly protected the voice of natural persons by referring to the form of applying the right of portrait. As a personal right, sound has the attribute of personal exclusivity. Any natural person's voice should be protected by law. Authorization of audio recordings does not mean authorization of AI based sound. Without the permission of the rights holder, unauthorized use or permission of others to use the sound in audio recordings constitutes infringement. With the widespread application of AI technology in various fields, disputes caused by AI technology are increasing day by day. On the China Judgments Online, reporters found many cases of infringement of the right to network dissemination of works, including keywords such as the use of voice AI. Regarding this, Lei Jiamao, a lawyer from Hebei Hounuo Law Firm, stated: "It is illegal for individuals to artificially transform others' voices into AI without their permission, which violates the legitimate rights and interests of others. Rights holders have the right to protect their rights in accordance with the law. According to Article 1023 of the Civil Code, for the protection of natural person voices, relevant provisions on the protection of portrait rights can be followed, and infringers can be required to stop infringing, apologize, and compensate for losses." Governance should take multiple measures and address both the symptoms and root causes. Currently, compared with AI face changing videos, AI synthesized speech is more difficult to distinguish and has a lower threshold. A phone call with a familiar voice may be a scam. On February 21st, the Cyberspace Administration of China released the key points of the 2025 "Clear and Bright" series of special actions to rectify the chaos of AI technology abuse, highlighting AI technology management and information content management, strengthening the generation and synthesis of content identification, cracking down on the use of AI technology to generate and publish false information, implementing network navy behavior, and regulating the network ecology of AI applications. Voice, as a personality right, has personal specificity. China has made certain provisions for the protection of sound rights within the existing legal framework Lei Jiamao said, "In the process of AI generating sound, there are multiple links involved, such as the collection and use of sound samples, the development and application of algorithms, and each link may involve the rights and interests of different parties. We need to further improve the legal system, refine the legal provisions of AI cloning technology, clarify the definition and responsibility of infringement, and strengthen law enforcement efforts Experts suggest that multiple measures should be taken to address the chaos of AI "stealing" sound and to address both the symptoms and root causes. At the platform level, it is necessary to effectively fulfill regulatory responsibilities, establish and improve the review and infringement reporting mechanism for AI voices, and timely detect and intercept harmful content such as infringement, falsehood, and illegality. Relevant departments should continue to increase their efforts to crack down on illegal and criminal activities such as fraud using AI technology, and form a more comprehensive and normalized governance mechanism. In addition, Lei Jiamao pointed out that in the era of increasingly popular AI technology, individuals must also strengthen their awareness of protecting personal biometric information, especially sound. Do not easily believe or spread unverified AI generated voice information to prevent being misled or involved in unnecessary risks. (New Society)