The "ai modified" video was created by a human machine
Recently, a video entitled "zhuge zhuge's live feed" was distributed on some web platforms. In the video, the turban's "zhuge liang" sells goods to the camera with modern words, and the screen is full of comments from liu bei's "semai". These “magic” videos, produced through ai, are often distorted and tautologically adapted to the classic literary images and historical narratives, and are disseminated through web platforms, with negative effects on youth groups。
In an effort to address the phenomenon of “ai transformation”, the national radio and television authority has been conducting a one-month nationwide campaign since 1 january 2026, requiring the internet-based audio-visual platform to carry out its primary responsibilities, strengthen content clearance, resolutely clean up irregularities, dispose of prominent accounts, reverse the negative trend in the spread of “ai transformation” videos and create a good cyberspace for the healthy development of adolescents。
"ai modified" video's online comment. Figure
It's a disturbing phenomenon
A number of parents interviewed by the chinese newspaper china-china youth network expressed concern about the phenomenon of “ai magic”. According to wang manman, a beijing parent (alias), her 10-year-old son recently used to watch a video called “funny history man”: “although the names of the characters in the video are familiar, the behaviour and lines are completely incorrect, the children sometimes take them seriously and they test us with them”. She feared that this situation would affect children's knowledge of history。
Shenzhen's parents found that their daughters in primary school were afraid at night, after they had seen some “black-style” cartoons to make short films, asking them “why the lovely cartoon figures are bleeding”。
In the comment section of some of the charades, "ai modified" videos, there were comments from netizens saying, "it's new for the first time, it's boring to see more." a number of netizens have expressed their dissatisfaction with the excessive “ai modification”, arguing that such content “has no meaning” and “disturbs”。
There are practitioners who reveal the mechanisms through which such content can be disseminated: these novel, unusual and even uncomfortable “ai-modified” sections often capture the eye in the first place, trigger the mechanism for recommending algorithms and thus gain flow. This has been used as a shortcut by some producers to focus on creating contrasts and pursuing sensory stimulation。
"how can you come?"
It was no accident that those elements had been rededicated, and the production and dissemination of the “ai transformation” video had resulted in a complete operating system。
An ai-generated video, adapted from " the legend " , was seen on the web platform by media publisher kim woo (alias), whose content contrasted with the original play and obtained very high interactive data. This led him to realize that there might be market opportunities for such content and then to start imitating production。
Kim woo described the need for producers to obtain raw video material from their video productions, and then to use the ai tool with the "facebook video" function to input static images into specific text instructions, for example, by requiring characters in the picture to perform a modern act that did not fit the original scene, thereby generating a highly dynamic and inverse short film。
The process would be more complex if the person were to “speak” a new line, combining voice cloning with oral synchronization, first copying the actor's tone to produce new voice and then adjusting the person's mouth to match it. Finally, the resulting video clips, audio effects and subtitles are synthesized by editing software。
Kim woo told reporters that the core of operating the "ai modified" account was "hot spots." for example, he said, “we're going to move the same idea `to the classics of `red house dreams', `lindayu', `guan yu', `tai yu', `dai yu', `tai yu' and `guang yu'.”
The key to “ai magic” is to create a strong sense of opportunism and new wonders, such as the use of modern objects by impersonators or behaviour that is completely incompatible with their original identity. In his view, such extreme contrasts were an effective means of obtaining instantaneous attention on the web platform。
The ultimate goal driving the entire process is the realization of flows. According to kim woo, the most basic method of realization is through the distribution of content through multiple platforms, relying on broadcast volumes to obtain platform subsidies. When the number of account fans increases to a certain size, commercial advertising orders may be received。
Moreover, teaching is a means of making money. Xu qing has established a paid community of hundreds of people who teach the operation in a standard set of processes, with corresponding tool software from material acquisition, ai generation, audio processing and later synthesis。
Video creators' pursuit of flow realization is an important driver of how such content slides to “how to get out”. “the core of ‘ai’ is not to create, but to do everything in its power to make the flow fast into money.” xu zing, speaking out。
It's time for "ai" to govern
Industry believes that, as the governance of ai generation content advances, the platform is not only a “pathway” for information distribution, but should be seen more as a “first responsible” and “core governing body” of web content ecology, with clear managerial responsibility for the safety and orientation of the content disseminated。
At the end of 2024, a verdict was handed down in the first ai vincent video dispute nationwide. The hunan provincial people's court of the chaifu district of hunan city found that the illegal dissemination of the short video clip of the jubilee by a network company through its development of the “ai-one-strike” function constituted a violation, ordered it to compensate the plaintiff for the economic losses and reasonable expenses of a technology company of the companion of the jubilee totalling 800,000 yuan and immediately to cease the violation. The changsha intermediate people's court upheld the judgement in second instance。
The changsha people's court held that in the case in question, a network company, without the authorization of the right holder, had provided users with 3 to 7 seconds of " year of the jubilee " , by means of an ai function of a software, and that the material sections of the work in question were part of the audiovisual work and were original and should be protected by law, and that the network company had uploaded the footage onto the server so that it could be accessed by the public on its own initiative, and that the act fell under the control of the right to disseminate information networks. As a provider of ai services, the network company has failed to meet its due diligence obligations and has not established an effective ipr abuse risk prevention mechanism. The court found that the defendant company had made a clear mistake in spreading the part of the jubilee through the ai function, thereby violating the plaintiff's right to disseminate information networks。
On 2 february, the national radio and television directorate issued a special progress report on the management of the “ai magic” video, noting that since the launch of the special governance exercise, the main web-based audio-visual platforms have effectively discharged their primary responsibility to enhance screening and disposal, to secure the implementation of the overhaul, and to identify and identify a number of typical cases of violations and violations to ensure effective governance. During the dedicated governance period, a cumulative amount of 23,000 videos of related violations were cleared and over 100 account numbers were disposed of。
It was reported that “ai transformation” video governance will be continued, with the construction of a normal, institutionalized long-lasting mechanism, adherence to both the symptoms and the treatment, and oversight of web-based audio-visual platforms to further strengthen corporate responsibility and build a healthy audio-visual space on an ongoing basis。
Relevant web platforms issue governance reports. According to a special summary report on governance issued by the platform on 1 february, the focus of governance is on a number of aspects, including the prohibition of the falsification of classic works such as four famous books, history and revolutions, and english model stories; the clean-up of content produced and disseminated through ai, which contains tacky, violent, poaching or unorthodox stereotypes; the prohibition of tampering with chinese civilizational symbols, confusing history and misleading cultural identities; and the complete clean-up of irregularities such as terror, hypocritical and so on, which are based on a classic animation image, and the strengthening of the protection of minors。
The platform indicated that specific governance requirements would be implemented on an ongoing basis, that it would improve its ability to identify and dispose of ai-generated irregularities, that it would deal with violations and accounts, and that it would call on a wide range of users to report violations and maintain a clear and orderly online community environment。
Be wary of the cosmopolitan confusion caused by "ai"
“lindayu” plays the "little willow" and “kwan yu” is a pair of fake leather shoes on the way to power ... In recent days, some of the absurd “ai modifications” have been transmitted on short video platforms, raising public concerns about cultural heritage, legal boundaries and technical ethics. A reporter from the china-china youth network interviewed a number of legal experts to provide an in-depth analysis of the legal risks and effective governance paths behind the phenomenon of “ai transformation”。
"ai" is not a simple "characterism" and the underlying legal risks are multilayered and cross-cutting. "'ai magic' is about two major kinds of work." the director-general of the institute of intellectual property law of the chinese university of political science and law continues to point out that a category of tampering with classic works that have entered the public domain, such as literary works, may infringe on the author's right to protect the integrity of his work, i. E. The right to prohibit others from misrepresenting or altering his work. “even after the expiry of the original copyright protection period, his original role image and the spiritual core are protected by law.”
He explained, for example, the image of guan yu, a historical figure whose story was public material, but the author's classic image of gwan yu, in the trinational presentation, was an original expression protected by law. By placing this image in a mythical scenario, "ai" distorts the image of the art and may constitute an infringement。
The other category is the “magic transformation” of works that are still under copyright protection, such as contemporary cinemas, with more complex legal issues. It was maintained that such “magic” acts may also violate the rights of authors to reproduction, to adaptation, to dissemination of information networks, and to the personal integrity of their works。
According to zhu zhu zhu, an associate professor at the chinese university of political science and law, the practice of “magistrates” transcends legal boundaries and often goes beyond the realm of copyright and may encroach on the realm of personal rights. “if the content of `magic' uses the portraits, voices of actors in the film, it is suspected of violating the rights of actors to portraits, voices; even if an actor dies, his close relatives have the right to seek legal protection.” he stressed that if the image of a martyr was tampered with in bad faith, the law could violate the law on protection of heroes and martyrs, to the detriment of the public interest, the prosecution authorities could institute public interest actions in accordance with the law。
It is worth noting that where “magic” acts involve commercial attributes, the legal regulation is stricter. “the creation of `magics' for the purpose of diversion and realization, using the well-known image of others, may constitute undue competition.”。
Zhongqing, a reporter from china, noted that, unlike traditional malfeasance, the content of "ai" is more authentic: for example, the "lindayu" is so detached, so it's perfect for the mouth, and the "gwan yu" speech is almost false. The iterative upgrading of technology makes it hard to see some of the bad stuff。
“this type of content generated by new technologies is more hidden, in some cases seemingly authentic and authoritative, and more misleading to minors.” the executive director of the beijing centre for legal aid and research on adolescents noted in xuquin that the shape of the content generated by ai is variable, that it is difficult to identify effectively with traditional keyword filters, and that such content is produced at a low threshold and, once recommended by algorithms, can easily spread virally。
“some minors are unable to distinguish between fictional boundaries and reality.” according to xuquin, many teenagers have not read the original "red house dream" or the 87th edition of the classic tv show, and their first knowledge of "lindayu" may be that of the "ai" version of "live willows," which could lead to deep cognitive confusion。
Zhu zhu agrees that “when the history of the revolution is played out and deconstructed, the cultural memory and historical reality of a people's common heritage will be eroded in the invisible. This is by no means a matter of alarm, but a serious question of the basis of cultural confidence.”
According to xuquin, there is a greater need to be vigilant about the “ai cult”, which uses the horrors and violence generated by ai to disguise it as an animated transmission by children. “the low threshold for the production of such content and its rapid spread can cause direct damage to the mental health of minors.”
Traditional governance frameworks face challenges in the face of massive ai violations. On 19 january, the central department for the defence of young people's rights and adolescents of the ycyc, in collaboration with the team of experts of the beijing centre for legal aid and research for young people, published the manual for the use of ai science, which aims to guide young people to improve their skills in the use of ai, to use it in a scientific and orderly manner, and to protect themselves from drowning, to enhance their ability to learn independently and to return to real social contact。
The regulatory level is working continuously. At the beginning of this year, the national general directorate of radio and television carried out a special campaign focusing on the “ai magic” content, which is based on four major titles, historical, revolutionary and other relevant television plays. In september 2025, departments such as the national network office jointly developed the manual for the marking of artificial intelligence for the production of synthetic content, which requires that the content generated by ai be made visible by force。
In interviews with journalists, experts agreed that the platform should no longer use the “technically neutral” and passive “notification-delete” rules as a shield. She stressed the need for the platform to assume its primary responsibility for matching its technical capacity with the establishment of an ai content identification mechanism, with a focus on monitoring high-risk account numbers and content, while opening channels for reporting and improving fast disposal channels。
The successor proposes a combination of fists to improve the systematic governance of the platform: the platform can proactively obtain a classic ip mandate to provide users with compliance-based, eco-regulating materials, while exploring the pattern of share of proceeds; optimize algorithms to limit and reduce the flow of low-quality, malfeasance, tampering with historical content and to cut off the chain of benefits from which it flows; and build the creator's ai copyright credit file, with disciplinary measures such as seals and bans on re-registration of frequent violators, to increase the cost of violations。
According to chu, governance of "ai magic" requires “technologies against technology” and the development of a more intelligent ai identification system that should address mass content audits. The necessary manual review chain must be maintained to ensure the accuracy of value judgements in response to special elements of major history and ingenuity. “second creation is an important form of cultural innovation and classical heritage and cannot stifle healthy creativity by governing `magic'.” chu stressed。
According to chu, the current legal framework is largely complete, with corresponding provisions in the law on the protection of heroic martyrs, the civil code, etc. “the key is the full and effective application of the law enforcement and judicial chain.” he argued that the core of governance “ai magic” was to find a dynamic and sustainable balance between encouraging innovation and protecting classics, technological freedom and legal norms, personal expression and the public interest。
A continuation of the proposal to build a participatory system of governance among regulators, platforms, industry organizations and users could explore formal channels of authorization for the creation of classic cultural symbols, drawing on the collective management model of copyright; at the same time, future legislation would further refine the regulation of the application of “deep synthesis” techniques, establish a higher level of protection for content related to classic cultural symbols, historical figures, etc., and introduce stricter punitive compensation mechanisms for malicious “ai modifications” for profit。
“only within a well-defined legal framework, and guided by technology towards good will, ai can truly be an instrument for enabling cultural prosperity.” chu said。
