In order to promote the construction of China's first full life cycle compliance standard for generative AI data, the Drafting Opinion Collection Meeting for the Group Standard for Compliance Guidelines for Generative AI Data Applications, hosted by the National Center for Industrial Information Security Development and Research (NCISDAR), was held in Beijing on December 28, 2023. The purpose of the meeting was to promote exchanges among all parties and to further enhance the scientific and practicality of the group standard. Mr. Wang Fei, Senior Partner of Beijing King&Capital Law Firm, participated in the meeting and delivered a speech.
As a representative of the drafter, Mr. Wang Fei put forward three suggestions on preventing training data from infringing intellectual property rights:
First, the consideration of preventing intellectual property infringement by training data should also take into account the principle of promoting the development and utilization of data. The data carrying intellectual property information is different from the intellectual property itself, and the two are not subject to the same legal regulation. The training data is only used to carry intellectual property information, and in this case, it can be divided into two cases, one of which is that the data itself constitutes intellectual property, and the other does not, and from the point of view of intellectual property rights, I personally believe that it can be used freely. If it constitutes intellectual property, express authorization should be obtained.
Secondly, the risk control of intellectual property infringement of training data should adhere to the principle of dynamic and full chain cycle verification. From the source of data, data processing multiple links multi-level control. For example, for data processing that may involve personality attributes, prior express individual consent is necessary.
Third, in the process of training data use, whether to pay attention to the destruction of the integrity of intellectual property rights and other subdivided issues is also an urgent response to the issue.
Finally, Wang Fei lawyer said, how to design the content can make the standard more vitality, in its statutory cycle to play a role in guiding the evaluation of standardizing the specific operation of the industry, intellectual property issues and generative artificial intelligence natural intersection of close, at the same time, in order to give the industry development and legal practice to leave space, it is recommended to use the enumeration plus the bottom of the way, set up a special article of intellectual property rights for the guidance of the clear.
It is reported that "Generative Artificial Intelligence Data Application Compliance Guide" is a group standard focusing on the full life cycle of generative artificial intelligence data compliance application, which will make up for the current gaps in national and industry standards in the field, with the China Electronic Chamber of Commerce (CECC) acting as the project management unit, the National Research Center for the Development of Industrial Information Security (the First Institute of Electronic Research of the Ministry of Industry and Information Technology) taking the lead in the drafting process, and the Zhihe Standard Center in the organization and preparation of the standard. It is a group standard focusing on the full life cycle compliance application of generative AI data, which will fill the gap of current national and industry standards in this field. Meanwhile, it is committed to helping enterprises better apply data in operation and R&D, enhancing their ability to monitor data risks, and contributing to the high-quality development of the service industry.
