news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Reporter of China Business Network: Ke Yang Editor of China Business Network: Wen Duo

On August 16, Shanghai Mita Network Technology Co., Ltd. (hereinafter referred to as Mita AI) issued a statement saying that it had received a 28-page infringement notification letter from China Academic Journal (CD-ROM Edition) Electronic Magazine Co., Ltd. (hereinafter referred to as CNKI).

On the afternoon of the 16th, MiTa AI told the reporter from the "Daily Economic News" that the subsequent provision of services will not be affected at all, and many databases have actively expressed their willingness to cooperate with it.

As the competition in AI search begins, new entrants have entered the segmented tracks to seek differentiated competition. The display of academic search results has always been the differentiated competitive advantage of "Mita AI Search". Whether this dispute with CNKI involves infringement may affect the future layout of Mita AI in academic search.

An industry insider said in an interview that in terms of copyright licensing and cooperation, AI search service providers can consider building a normalized cooperative relationship with large data resource websites or content copyright holders based on their future business positioning and scale, obtain complete licensing through cooperation agreements, use their corporate technological advantages to obtain cooperation benefits, share profits, and better solve infringement problems at the root.

Industry insiders: Whether there is infringement depends on these factors

According to the article published by Mita AI, HowNet said that it was discovered that Mita AI provided users with academic literature titles and abstracts of HowNet through "Mita AI Search" and "Mita AI Search App", and the amount was huge. HowNet believes that this behavior is without the permission of HowNet and seriously infringes on the legitimate rights and interests of HowNet. It also said that it does not want China HowNet to be searched by Mita Technology and requires Mita AI to immediately disconnect the link from the search results to the HowNet website.

"Without search, there is no research. 'Knowledge discovery' is the first step in the knowledge dissemination chain." Mita AI said that from now on, "Mita AI Search" will no longer include the title and abstract data of CNKI documents, but will instead include the title and abstract data of documents from other authoritative knowledge bases in Chinese and English.

On August 16, a reporter from the Daily Economic News opened the Mita search and found that a jump link for the above statement was added to the "Mita AI Search" webpage. The reporter used the "Mita AI Search" and found that the CNKI link was no longer displayed in its references, and most of the links pointed to other domestic and foreign websites.

On the afternoon of the 16th, MiTa AI responded to the reporter of "Daily Economic News" on WeChat and stated that the subsequent provision of services will not be affected at all, and many databases have proactively expressed their willingness to cooperate with MiTa.

Regarding whether MiTa AI constitutes infringement, Chen Shuang, senior partner of Beijing Yinghe Law Firm and Chinese expert member of the International Data Management Association DAMA, said in an interview with the "Daily Economic News" reporter that whether the infringement claimed by HowNet is established depends first on whether HowNet has the rights to the data resources it claims, what rights it has, and whether MiTa AI's use of the data resources is improper and thus constitutes infringement.

Chen Shuang believes that usually the copyright of the abstract belongs to the author of the paper. If HowNet has actually obtained the full authorization of the copyright holder through the copyright agreement, that is, it not only has the right to include, but also has the right to authorize the use of the document to others, sue on behalf of the copyright holder, and other complete rights, similar to the role of "document manager", then the AI ​​search product may infringe the right of information network dissemination by directly copying and displaying it to users. Bibliographic data usually includes the title, author, document source, document time, etc., and generally does not constitute a work. However, if HowNet can provide the copyright registration certificate for the document abstracts and bibliographic databases it claims, or there is evidence to prove that it has made original labor input in the process of collating, compiling, and generating these data, rather than simply piling up various elements and accepting them all, then there is a possibility that it can claim property rights to the data resources from Mita AI.

In addition, Chen Shuang also said that the underlying technology of AI search products is usually mainly crawler technology. Whether the "crawling" behavior of AI search products is legitimate, in addition to determining the source of the crawled data and the compliance of the content, it is also necessary to combine whether it complies with the crawler agreement, user agreement, and platform rules of the crawled website, whether it has taken measures to break through or bypass the website's anti-crawling technology, and whether the use and display methods constitute a substantial replacement for the crawled website. At the same time, from the perspective of public interest, the court will also consider the balance between the boundaries of data rights protection and the sharing and circulation of academic resources to make a comprehensive judgment.

It is recommended that search service providers cooperate with large data resource websites, etc.

MiTa AI, the company was founded in April 2018 and focuses on the field of artificial intelligence.

In March this year, "Mita AI Search" was launched. In early August, media reported that Mita AI recently completed a new round of financing of more than 100 million yuan, led by Ant Group and followed by Lightspeed Photosynthesis. The post-investment valuation of this round reached 150 million US dollars.

Recently,OpenAIThe official announcement of joining the AI ​​search field has made the already crowded AI search track even more lively. In foreign countries, there are old companies such as Microsoft Bing Search (Bing), Google AI Overviews, andPerplexityEmerging AI search companies such as Mita AI and Kunlun Wanwei have continued to increase their investment in AI search. In China, traditional search giants such as Baidu and "search newcomer" Quark have also added AI functions to their products.

As competition in the AI ​​search market becomes increasingly fierce, many players are beginning to seek differentiated advantages in segmented tracks. On July 26, user experience expert and columnist Du Zhao said in an interview with a reporter from the Daily Economic News that it is difficult for new entrants to break through the capabilities of search engines in terms of data resources, model capabilities, index libraries, etc., which requires huge costs. Therefore, the differentiation of different AI search products may be reflected in segmented fields, and it is more difficult to differentiate general search.

However, as MiTa AI received the infringement notification letter from CNKI, future AI search players will also need to start thinking about how to deal with possible infringement risks.

In this regard, Chen Shuang said that domestic service providers similar to Mita AI should meet the compliance requirements of the Interim Measures for the Administration of Generative Artificial Intelligence Services when providing generative retrieval services. According to Article 7 of the Interim Measures for the Administration of Generative Artificial Intelligence Services, artificial intelligence service providers "shall carry out pre-training, optimization training and other training data processing activities in accordance with the law" and comply with the provisions of "where intellectual property rights are involved, they shall not infringe on the intellectual property rights enjoyed by others in accordance with the law".

Chen Shuang suggested that the crawler technical means and procedures should be standardized, the crawler protocols and rules of the crawled website should be followed, and data should be crawled by avoiding any web page technical restrictions or violating the website robots protocol. The possible impact on the performance and function of network services should be evaluated in advance to avoid interfering with the normal function of the crawled website network services or hindering the normal operation of computer information systems. For similar data training on large resource platforms, a content review mechanism should be established, and an internal review process of "technical means + manual assistance" should be formulated, such as reviewing the platform user agreement/platform rules and confirming the status of the crawled platform's data rights.

In addition, Chen Shuang also mentioned that the platform should set reasonable usage restrictions. For academic resources protected by copyright, the platform should set reasonable usage restrictions in the user agreement, such as prohibiting or restricting users from copying, distributing or using relevant search result links for commercial purposes. In terms of copyright authorization and cooperation, service providers like Mita AI can consider building a normalized cooperative relationship with large data resource websites or content copyright holders based on future business positioning and scale, obtain complete authorization licenses through cooperation agreements, obtain cooperation discounts with corporate technological advantages, share profits, and better solve infringement problems from the root. Rapid review and response processing of user or right holder complaints. If it is indeed an infringement, the relevant infringing content will be removed from the shelves in a timely manner.

Daily Economic News