Risks and Governance Paths of Generative Artificial Intelligence in Mainstream Media Communication / Риски и пути управления генеративным искусственным интеллектом в основных средствах массовой коммуникации

Ма Фэй ORCID: 0009-0002-5688-1348 аспирант; филологический факультет; Российский университет дружбы народов им. П. Лумумбы 117198, Россия, г. Москва, ул. Миклухо-Маклая, 6 Ma Fei Postgraduate student; Faculty of Philology; RUDN University Miklukho-Maklaya str., 6, Moscow, 117198, Russia 1042238104@pfur.ru Булгарова Белла Ахмедовна ORCID: 0000-0001-6005-2505 кандидат филологических наук доцент; кафедра массовых коммуникаций; Российский университет дружбы народов им. П. Лумумбы 117198, Россия, г. Москва, ул. Миклухо-Маклая, 6 Bulgarova Bella Akhmedovna PhD in Philology Associate Professor; Department of Mass Communications; RUDN University Miklukho-Maklaya str., 6, Moscow, 117198, Russia bulgarova-ba@rudn.ru Другие публикации этого автора     Чэнь Фэнлань ORCID: 0009-0007-2297-6126 аспирант; филологический факультет; Российский университет дружбы народов им. П. Лумумбы 117198, Россия, г. Москва, ул. Миклухо-Маклая, 6 Chen Fenglan Postgraduate student; Faculty of Philology; RUDN University Miklukho-Maklaya str., 6, Moscow, 117198, Russia 1042238094@rudn.ru Другие публикации этого автора

Мондал Папия ORCID: 0009-0006-0337-7059 аспирант; филологический факультет; Российский университет дружбы народов им. П. Лумумбы 117198, Россия, г. Москва, ул. Миклухо-Маклая, 6 Mondal Papiya Postgraduate student; Faculty of Philology; RUDN University Miklukho-Maklaya str., 6, Moscow, 117198, Russia 1042235023@pfur.ru

Scientific and technological development has increasingly become an important reference dimension for evaluating the competitive ability between countries, and the leapfrog breakout and advanced development of generative artificial intelligence represented by ChatGPT, Sora, etc. has once again highlighted the enormous development potential brought about by scientific and technological innovation. At the same time, it also highlights the urgency and importance of accelerating the standardized governance of generative AI. The intelligent service experience brought about by the advantages of generative artificial intelligence in human-computer interaction and human-computer collaboration not only improves social productivity and social intelligence, but also gradually becomes an important technology driving force that pushes human society forward. In the field of media, the development and application of generative artificial intelligence has promoted a huge change in the production and communication of mainstream media content. As a new type of knowledge-generating medium, it is accelerating the rewriting of the content production and communication pattern in the media field. However, at the same time, the leapfrog progressive development and broad-spectrum application of generative AI have also extended the AI risk governance boundary and governance complexity. Therefore, based on the existing research results of generative AI in the media field, this paper attempts to systematically analyze and study the risks of generative AI in mainstream media communication, and puts forward the governance path to cope with the risks.

1. Origins: A Review of Generative Artificial Intelligence Research in the Media Field

Artificial intelligence is an interdisciplinary and comprehensive concept, which was initially proposed at an academic conference held at Dartmouth College in the United States in 1956, mainly referring to the creation of intelligent machines or programs. Since the birth and rapid development of artificial intelligence to date, artificial intelligence technology has experienced three waves, based on the nature of the technology has gradually shifted from “logical reasoning” to “knowledge engineering” and “machine learning” ^{[1, p. 26-50, 155]}. And supported by a variety of digital technologies such as arithmetic power, algorithms and big data together, the new generation of artificial intelligence is driving human society into the fourth industrial revolution ^{[2, p. 139-159]}. With the leaping and iterative development of AI technology, generative AI represented by ChatGPT, Sora, etc. has ushered in a leaping progression since 2022, inspiring profound social changes in various fields with the support of the three key technologies of pre-training, large modeling, and generative ^{[3, p. 81-90]}. So that people have witnessed a huge leap in AI technology in terms of fidelity and functional dimensions, and the initial prototype of generalized AI ^{[4, p. 147-160]}. Its powerful self-learning, self-supervision and evolution capabilities have gradually opened up a new era of general artificial intelligence. This signals that mankind will enter a more complex and risky intelligent system.

With the rapid iterative development of a variety of digital technologies, generative artificial intelligence is gradually embedded in the field of media communication, greatly enhancing the efficiency of content production and communication. The content production mode of the media is gradually changing from professional content production, user content production to artificial intelligence production. Generative artificial intelligence is accelerating the transformation and upgrading of digitization and intelligence in the media field. Back in 2015, Tencent Finance released writing robots such as Dreamwriter that caused a stir in the industry. However, this only improved content productivity to an incremental degree, and did not play an essential and revolutionary change. The iterative development of generative artificial intelligence breaks through the inherent tool attributes of technology, and the active exploration and application in the media field, content production and communication driven by generative artificial intelligence development has gradually become the norm and mainstream ^{[5, p. 29-35]}. The development of generative artificial intelligence has realized the leap to the subject attribute, changed the traditional pattern of communication subject, and realized the change from “human” to “human” and “machine” coexistence ^{[6, p. 62-69.]}.

Fundamentally, the rapid iterative development of generative AI technology further promotes the intelligent development trend of media communication, triggering disruptive changes in the mainstream media communication landscape. The widespread application of generative AI in the media field, represented by Chat GPT, has become an important milestone in the intelligent upgrading of media. Relying on its powerful comprehension ability, it has demonstrated significant advantages in text creation, video image generation and other aspects ^{[7, p. 32-36]}. On the one hand, it fits the current intelligent development trend of information communication. On the other hand there are many risks associated with the widespread application of generative artificial intelligence, such as the risk of data privacy and security, and the risk of authenticity and accuracy caused by false information. Overall, scholars generally hold two attitudes towards the application of generative artificial intelligence in the media field ^{[8, p. 37-40, 59.]}. One attitude is that generative AI is a super tool that will bring about profound changes to the order of communication, media ecology and so on. Another attitude is that it is only an auxiliary tool in the field of media communication and will not change the basic operating principles of the media industry.

In summary, it can be seen that the powerful functional characteristics of generative artificial intelligence are widely used in the media field, which improves the efficiency of content production and communication and enhances the user's intelligent service experience. At the same time, it also brings a lot of risks and challenges, further aggravating the hidden and diversified nature of risk prevention. Therefore, it is necessary for the national government, media platforms, individual users and other multiple subjects to participate in governance.

2. Challenges: Risks of Generative Artificial Intelligence in Mainstream Media Communication

The rapid iterative development of generative artificial intelligence and its wide application in mainstream media communication have injected new variables into the ecology of communication and public opinion. It not only expands the breadth and depth of the application field, but also extends the AI risk governance boundary and governance complexity. Therefore, this part will specifically analyze the possible risk issues brought by generative AI in mainstream media communication from the perspectives of data privacy and security risk, the risk of manipulating public opinion with false information, and the challenge of media trust.

2.1 Data Privacy Security Risks

Ian Ayres has mentioned that “big data analytics reduces the space for effective privacy, and people increasingly live in a world where they can't hide who they are, what they've done and what they'll do ^[9].” In the development of generative AI, data is the basis of its operation, an important resource for the development of AI, and to a certain extent, it is also seen as the core of intelligent communication ^{[10, p. 42- 56]}. In the process of mainstream media communication, the user's application of generative artificial intelligence has data privacy security risks in the process of data collection, storage and transmission. For example, when users are acquiring information through intelligent software, the application system often requests access to the user's personal privacy data permissions, and in this process, if personal information is excessively collected or collected in an opaque manner, the user's data privacy will be leaked and other security risks. Meanwhile, in the process of human-computer interaction, when the user and the generative AI carry out human-computer interaction questions and answers, the user needs to give up part of the personal data information in order to obtain the “private customized” intelligent services. After obtaining the relevant data, the generative AI generates a user profile through data aggregation and analysis, thus obtaining a high-quality human-computer interaction experience. Of course, in the early stage of the use of generative AI, most users will use it cautiously due to data security considerations, but in the process of human-computer interaction, generative AI can still obtain various types of data such as the user's linguistic style characteristics, personal data information, current focus, and even identity in the process of information exchange. According to the data in the “Data Leakage Risk Landscape Report for the First Half of 2024”, the number of data leakage incidents across the network reached 16,011 in the first half of 2024 alone, an increase of 59.58% compared to the data in the second half of 2023. This growth data also shows that data privacy security risk is an issue that requires urgent attention and increased vigilance.

2.2 Risk of Manipulation of Public Opinion through Disinformation

The embedding of generative AI in mainstream media has improved the efficiency of content production and communication and enhanced the user experience to a certain extent, but at the same time, it also provides a natural breeding ground for the breeding and proliferation of false information. In particular, the application of cutting-edge technologies represented by deep fake makes the content forms of generated disinformation more diverse and more difficult to recognize. Disinformation is usually referred to by the concept of “Disinformation”, which refers to information that contains false or misleading content. The expressions “information fog” ^{[11, p.74-81, 96]}, “misleading information” ^{[12, p. 1-10]} and “misinformation” ^{[13, p. 102-113]} were then derived. The purpose of communicating disinformation is to mislead the public through misinformation in order to achieve a certain goal. From the perspective of technology application, the development of generative artificial intelligence has further lowered the threshold of false information production and communication, and a large amount of false information has appeared in the media communication field showing the phenomenon of “information fog”. At the end of 2023, the U.S. research organization released a report showing that fake news sites created using AI agents surged from 49 to more than 600 in seven months ^[14]. While the application of artificial intelligence is increasing the efficiency of content production, it is also accelerating to become a communicator of false information and misinformation.

Deep fakes are alternative digital representations that have been processed or generated by artificial intelligence so as to be seen as real faked images and audio and video ^[15]. With the rapid development of artificial intelligence technology, the multimodal function expansion of Deep Fake Technology not only makes false information more confusing and difficult to distinguish between true and false, but also makes it become a “weapon” to manipulate public opinion through the creation of false information. For example, some external forces may use Deep Fake Technology to generate false information or discriminatory information, and accelerate its communication to the field of public opinion, so as to realize the manipulation of the public opinion field and influence the direction of public opinion. Once these confusing and false information is accepted by the public, it is very likely to affect the value judgment and cognitive position of the public towards the country, society and events. Therefore, in mainstream media communication, we should be alert to the risk of manipulating public opinion by generating false information using Deep Fake Technology.

2.3 Media Trust Challenges

Media trust is the degree of public confidence in media organizations and the information they provide. This relationship of trust is based on the authenticity, reliability and accuracy of media organizations ^{[16, p. 29-31]}. For mainstream media, especially in the process of rapid development of generative artificial intelligence and its wide application in mainstream media communication, the correlation between the public's trust in the media and the authenticity and reliability of the communication content is more obvious. In terms of authenticity, authenticity is the life of news, and the influence of generative AI to generate the authenticity of communication content is manifested on the one hand in the authenticity and reliability of the training data. The enhancement of characteristics such as intelligence and humanization of generative AI needs to be realized through data training, and most of the training data come from open source data, automated data collection and public data. To a certain extent, these training data contain false information, misleading and biased contents. In the process of human-computer interaction and automated generation, it is possible to produce “artificial intelligence illusion”, generating some seemingly reasonable but inaccurate content, affecting the public's overall perception of the communication content, thus reducing the public's trust in the communication content of the mainstream media. On the other hand, it is manifested in the timeliness of the data. The content generated by generative AI automation is largely based on the static database contacted during data training, which cannot be updated or provide the latest news reports in a timely manner. This means that the content generated by mainstream media through generative AI may have a certain lag. This will, to a certain extent, affect the authenticity and accuracy of the communication content, which in turn affects the public's trust in the media. Therefore, for content that needs to be reported and communicated in a timely manner, it is also necessary to ensure the timeliness of the communicated content through real-time data collection by media editors. In summary, it can be seen that for the possible risks of authenticity and accuracy that may exist in the communication of generative AI in the mainstream media and affect the public's trust in the media, it is necessary to strengthen the fact checking, reduce the interference of misleading or biased information to the public, and build a trustworthy cognitive space for the public with authentic and timely communication content.

3. Governance: Paths to Generative AI Risk Governance in Mainstream Media Communication

With the rapid iterative development of generative AI technology and its gradual use in mainstream media communication, strengthening AI governance has become a general consensus in society. The state, media organizations, operation platforms, and the public play important roles in risk governance, so it is important to give full play to the initiative and mobility of each subject in risk governance, and to jointly promote the safe, reliable, and healthy and good development of generative AI in mainstream media communication from the levels of policy leadership, regulation and technology, global collaboration, and literacy enhancement.

3.1 Policy Leadership: Constructing and Standardizing the Direction of Governance and Strengthening Self-regulation in the Media Industry

From the perspective of risk governance, governments are primarily responsible for addressing the risks of generative AI, playing an important role as supervisors and risk prevention and governance. At the level of government policy leadership, more and more countries in the world are rapidly moving from watching and not knowing what to do about generative AI to introducing policies to lead the construction and standardization of governance. For example, in June 2023, the EU Parliament passed the Artificial Intelligence Bill, and the latest draft added provisions specifically regulating the safety risks of generative AI ^[17]. In July 2023, the Canadian Center for Cyber Security (CCCS) released a guide, Generative AI, which details possible security risks such as data privacy, algorithmic bias, and disinformation associated with the use of generative AI ^{[18, p. 48-53]}. In November 2022, China adopted the Administrative Provisions on Deep Synthesis of Internet Information Services to regulate the risk of generative AI content more comprehensively ^{[19, p. 147-160]}. In July 2023 China issued the Interim Measures for the Administration of Generative Artificial Intelligence Services to regulate the provision and use of generative AI services. From this, it can be seen that countries around the world are enhancing their risk assessment and control capabilities and standardizing the direction of generative AI governance through policy leadership. At the level of the media industry, as generative AI becomes more widely embedded in mainstream media communication, the media industry should continue to improve its internal self-regulatory mechanism, clarify the media industry's code of conduct when using generative AI technology, and encourage media organizations to clarify the role of AI technology in the entire process of content production and communication ^{[20, p. 29-31]}. Actively exploring the enhancement of fact-checking capabilities through generative artificial intelligence technology to improve the authenticity and accuracy of communication content.

3.2 Regulations and Technology: Improving Legal Norms and Data Management Processes

Legal regulation and technical statute are important links to prevent the risk of generative artificial intelligence in the field of media communication. Currently, generative artificial intelligence is still in the early stage of development, and as its application field in mainstream media communication gradually expands, timely and moderate legal norms and technical regulations are needed to promote the safe and healthy development of generative artificial intelligence on a safe track. First, improve the legal norms and update the concept of regulatory governance. The Interim Measures propose that the governance of generative AI should uphold the concepts of inclusive, prudent, and classified and graded regulation. Inclusive and prudent and classification and grading supervision has gradually become a new regulatory model recognized and accepted in the new industry because it is more in line with the regulatory thinking in the new economic form. As a rapidly developing new field and new form, this new regulatory model can minimize the interference of the emerging industry and maximize the regulatory benefits, and guide and regulate the safe and healthy development of generative artificial intelligence in the media field. For example, establish a perfect data classification and grading supervision mechanism, screen training data from the data source, improve the quality of training data and help release the value, and ensure the accuracy and objectivity of the output content in human-computer interaction and content generation. At the same time, we fulfill the obligation to protect the information and interaction records entered in human-computer interaction, so as to effectively and efficiently protect the security of personal data. Second, with the concept of good intelligence as the lead, the data management process is sound. With the iterative development of generative artificial intelligence, although the accuracy and richness of autonomous learning and generated content can be improved through massive data training, the continuous iteration of algorithmic technology has also led to more complexity in the hidden layer of technology. Therefore, the concept of good wisdom of good technology and good governance should be embedded in algorithmic technology to regulate the development of technology. On the one hand, internally, algorithmic supervision technology should be strengthened in the development of generative AI to realize effective internal code governance. On the other hand, a third party should be introduced to supervise the development and application of algorithms and establish a user feedback system to realize effective external supervision ^{[21, p. 21-30]}. In this way, we can realize the reduction of risks that may be caused by algorithms and other technologies from the inside out and led by the concept of Good Intelligence.

3.3 Global Collaboration: Strengthening Dialogue and Establishing Cooperation to Improve Global Artificial Intelligence Governance

The Global Initiative on Artificial Intelligence Governance calls on countries to strengthen information exchange and technical cooperation and to jointly address the risks posed by the rapid development of science and technology. The current world situation is evolving at an accelerated pace, and countries are clearly in competition and wariness of each other in terms of exploring exchanges and cooperation in the development and governance of artificial intelligence. And the risks posed by the iterative development of generative AI can no longer be faced and solved by any one country alone. As a global challenge, it requires the international community to strengthen cooperation and synergize to deal with it in terms of technological innovation, data resource construction and risk prevention. Therefore, strengthening communication and technical cooperation between countries around the world, accelerating the establishment of transnational communication and cooperation mechanisms, and improving global AI governance have become the way to deal with the security risks and challenges of generative AI. First, formulate international standards and safety norms. Countries should formulate standards and norms on aspects triggered by generative AI, such as data privacy and security protection, algorithm transparency, security ethics and morality, to ensure its safe and controllable development. Second, the media of various countries should strengthen dialogues and exchanges, establish a sense of sharing and co-construction, and build a transnational cooperation mechanism. Share experiences in technology research and development, data analysis, talent cultivation, etc., and exchange the latest developments and latest achievements. Promote the safe development of generative AI in the media field in the cohesion of consensus. Third, improve global AI governance. In the face of the risks brought about by the rapid iterative development of technology, mankind is in the same boat and shares the same fate. Therefore, perfecting global AI governance is precisely providing possibilities for solving the transnational governance problems lurking behind the iterative development of generative AI ^{[22, p. 147-160]}. Based on this, from the perspective of global collaboration, we should actively carry out international cooperation and global action, uphold the spirit of security cooperation of “common, safe, cooperative and sustainable”, improve the governance of AI, and promote the safe, reliable and healthy development of generative AI in various fields such as mainstream media.

3.4 Literacy Enhancement: Strengthening Intelligent Cognition and Improving People's Digital Literacy

Paul Gilster, the founder of digital literacy theory, explains digital literacy as being the ability to understand and use polymorphic information available through the Internet ^[23]. Digital literacy is a combination of qualities and abilities such as information acquisition, interaction and sharing, ethics and morality in learning practices ^[24]. With the intelligent iteration of generative AI and its increasingly widespread application in mainstream media communication, intelligence is gradually becoming a new way of being characterized in the daily life of the people. This intelligent mimetic environment requires the public to improve their cognitive ability in terms of the information they receive. Therefore, improving people's digital literacy has become a key link and an inevitable trend to adapt to the evolution of the times and prevent risks. On the one hand, it is important to improve people's awareness of the application of intelligent products. The application of generative AI in mainstream media communication has improved the intelligent experience of the public, while the public knows little about the application of generative AI represented by ChatGPT. In the future, as the application of AI in mainstream media communication becomes more extensive and the experience more intelligent, it is more necessary for the public to have a more comprehensive understanding and knowledge of it. And to look at the intelligent change of mainstream media with a more open mind, to better adapt to the intelligent environment in the era of intelligent media ^{[25, p. 65-72]}. On the other hand, it is important to enhance security guidance and digital literacy. In the face of the new digital divide and “digital leviathan” brought about by generative AI, we should enhance the people's ability to recognize and identify false information, public opinion manipulation and other contents and behaviors, cultivate the ability to correctly obtain and judge information, as well as the awareness of ethical and social responsibility, develop higher-order thinking, and improve people's digital literacy.

4. Conclusion

“The Smart Age is an era of development driven by the convergence of multiple digital technologies such as big data and artificial intelligence ^{[26, p. 14-19]}”. With the emergence and rapid development of generative artificial intelligence, the concept of security has been continuously extended and expanded, and the concept of security is more and more based on cutting-edge technology and its prospective consideration, bringing a deeper level of change and reconstruction ^{[27, p. 106-116]}. Along with the leap-forward development of generative AI and its broad-spectrum application in mainstream media communication, human beings are also witnessing the profound transformation of AI technology in terms of high intelligence, deep interactivity, etc., and seeing broader and more realistic intelligent application scenarios, as well as driving the development of human social life in the direction of greater intelligence and personalization. Generative AI with powerful human-computer interaction, human-computer collaboration and other capabilities gradually opened a new era of general artificial intelligence. At the same time, the rapid iterative development of generative AI also indicates that human beings are gradually entering a high-risk society operated by an intelligent system that is more complex and full of security risks. Therefore, in the face of the leapfrog iterative development of generative AI, we should not only see the new kinetic energy it provides for mainstream media communication, but also provide an integrated and comprehensive governance program for the innovative development and ecological construction of generative AI under the guidance of the concept of the security concept and with the multidimensional and holographic view of safety, in terms of conceptual awareness, conceptual scales, and technological methods ^{[28, p. 109-118, 155-156]}. This will not only give full play to the technical effectiveness of generative AI, but also jointly promote the safe, reliable and healthy development of generative AI in mainstream media communication.