IT Home reported on December 20 that the China Audio-video and Digital Publishing Association today released the group standard of "Guidelines for the Application of Generative Artificial Intelligence Technology in the Publishing Industry".
In accordance with the relevant requirements of the "Provisions on the Management of Group Standards of the China Audio-video and Digital Publishing Association", the group standard "Guidelines for the Application of Generative Artificial Intelligence Technology in the Publishing Industry" is hereby approved for release after procedures such as project review, standard drafting, solicitation of opinions, and review by the expert group, and has passed the review of the Youth League Standards Committee. The standard number is T/CADPA47-2023, published on December 20, 2023, and will be implemented from January 20, 2024.
The original text of the "Guide to the Application of Generative Artificial Intelligence Technology in the Publishing Industry (Draft for Comments)" is attached by IT House:
A guide to the use of generative AI technology in the publishing industryrange
This document sets out the basic principles, main application scenarios and corresponding safeguards for the application of generative AI technology in the publishing industry. This document applies to services provided by publishers using generative AI technologies.Normative references
The content of the following documents constitutes an essential provision of this document by means of normative references in the text. For dated references, only the version corresponding to that date applies to this document, and for undated references, the latest version of the document (including all change orders) applies to this document.
- GB/T 35273—2020 Information Security Technology - Personal Information Security Specification
- CY/T 96—2013 e-book content terminology
Terms and Definitions
CY/T 96 and the following terms and definitions apply to this document.生成式人工智能技术 generative artificial intelligence technology
Natural language processing technology that automatically generates text, pictures, sounds, videos, codes, and other content based on algorithms, models, and large-scale corpora.出版机构 publishing institution
A professional body engaged in publishing activities.Service provider
Organizations or individuals that provide generative AI services to the public in mainland China in the form of interactive interfaces, programmable interfaces, etc.Basic principles
Orientation
Ensure that the content generated by AI technology is correctly oriented, complies with relevant national laws, regulations and ethical guidelines, and respects social values and moral standards, so as to promote the application of generative AI technology for good.Compliance
The data used for the training of generative AI technology is from legitimate sources and respects intellectual property rights. At the same time, relevant service providers do not abuse their dominant market position, comply with anti-monopoly laws and regulations, maintain a level playing field, and promote innovation and diversity.Privacy
Publishers and service providers shall protect personal privacy in the process of data collection, storage, use and sharing in accordance with GB/T 35273 and other relevant national regulations.safety
Publishers and service providers should provide relevant measures in the process of data collection, storage, use, processing, transmission, provision, and disclosure to ensure data security and controllability. At the same time, the service provider shall provide secure, stable and continuous services to ensure the normal use of the publisher and other users.transparency
In the process of applying generative AI technology, all participants, including publishers and service providers, should understand and clearly disclose the use of AI content generation, including the underlying data set, data sources, and data processing methods.Fairness
Carefully evaluate and review potential sources of data and content bias during training data selection, algorithm design, model generation and optimization, and use, while reducing cultural or linguistic inequities in AI-generated content.accuracy
The data sources for pre-training and optimized training using artificial intelligence technology should be accurate, the description should be objective, and the data should be diverse, so as to avoid the spread of false information and misleading content.Advancement
Publishers select service providers with advanced AI-generated content technologies and algorithmic models, that is, the service providers have leading algorithmic core indicators (such as parameter scale) and authoritative evaluation scores in the industry.applicability
The AI-generated content technologies and tools provided by service providers should meet the needs of publishers for content resource management and application. The content generated by AI generated by value-based publishing institutions is ideological, artistic, intellectual, logical and scientific.Main application scenarios
Information Collection
According to market demand or topic planning, use generative AI technology to select appropriate content for publication in terms of information search and information integration.Assist in creation
Utilize generative AI technology tools to help editors and writers generate draft content for creative writing.Assisted audits
Generative AI technology tools are used to conduct a preliminary review of whether the published content meets the requirements of the state for publication management, especially the ideology, laws and regulations, sensitive content, copyright, etc., and provide the corresponding review report and logo for manual review and confirmation.Intelligent editing and proofreading
Generative AI technology tools are used to carry out preliminary linguistic editing and processing, as well as word error proofreading, normative proofreading of terminology, and proofreading of style and formatting. Processing work in the form of typesetting, typography and other forms can also be done with the help of generative AI tools.Publicity
Use the content generated by artificial intelligence technology to carry out new media operations, the promotion of published content or the brand marketing activities of publishing institutions, so as to enhance the brand influence of publishing institutions and the reach, reading rate and influence of published content.Assist in decision-making
The data analysis capabilities of generative AI technology can identify industry trends and patterns, and guide publishing plans and strategies to aid publishing decisions.Knowledge Services
Using generative artificial intelligence technology, based on information collection, organizational analysis, and according to user needs and information environment, we can provide publishing content services that can effectively support knowledge application and innovation, and can indicate the necessary citation sources and references.Other scenarios
Publishers can use generative AI technology to assist in the design of publication covers, illustrations and other content, as well as the auxiliary translation and semantic analysis of publication content, as well as data collection, collation and analysis.Management mechanism
Publishers
In the process of applying generative AI technology and related tools, publishing institutions should comply with the "Regulations on the Administration of Publications", the "Provisions on the Administration of Online Publishing Services", and the organization's publishing business process management specifications, strengthen content screening, review, and personnel training, and do a good job of relevant content security and quality and safety management.Service Providers
Service providers need to establish corresponding management systems for the review of the mechanism of algorithms, the review of scientific and technological ethics, user registration, the review of information release, data security and personal information protection, and the emergency response to security incidents, and allocate professional personnel and technical support appropriate to the scale of their services.intellectual property
Publishers and service providers are required to make full and correct disclosures and declarations about the use of AI-generated content, and they shall not use the content of the relevant works in any way or in any form without the specific and explicit permission of the right holders and relevant stakeholders.security
Corpus security
Publishers and service providers should establish relevant blacklists of corpus sources, strengthen the security assessment of corpus from each source and the training and assessment of annotators, and ensure the security of corpus used for pre-training of generative AI technology.Model security
The basic model of generative AI technology used by service providers should be filed with the competent national authority, the user of the service provided and the use of third-party basic models should be disclosed, and the security analysis of user input information and the accuracy of the generated content should be strengthened to ensure the security of the model of the service provider.Data Security
When applying generative AI technology, effective encryption and control methods should be used for data processing, transmission, and access in accordance with relevant national regulations, to ensure data security and protect personal privacy.Cybersecurity
In accordance with relevant national regulations, configure corresponding network security measures such as firewalls, network monitoring tools, authentication measures, interface security, and intrusion detection.System security
When providing generative AI technology services, service providers need to comply with relevant national regulations and have secure system architecture design, operating system and hardware security guarantees, and applications to ensure the safe and smooth operation of the system.