In 2024, AI and intellectual property right issues were widely discussed in Japan, as in many other jurisdictions. The main recent developments in Japan are discussed below.

Report on AI and Copyright Issues (15 March 2024)

The Copyright Act of Japan (the “Copyright Act”) is administered by the Agency for Cultural Affairs of Japan (the “Cultural Agency”). The Cultural Council, a body established under the Cultural Agency to research and review matters relevant to the Cultural Agency, has a further subordinate body called the Copyright Subdivision, which researches and reviews matters related to copyright. On 30 June 2023, the Copyright Subdivision decided that its Legal System Subcommittee (the “Subcommittee”) would research and review certain AI and copyright issues, and the Subcommittee commenced its review on 26 July 2023. The Subcommittee published a preliminary draft report on those AI and copyright issues and requested public comment from mid-January to early February 2024. In response to this request, 24,938 public comments were submitted. After approximately seven months of research and review, taking into account the comments received from the public, the Subcommittee published its report, titled “Perspectives Regarding AI and Copyright” dated 15 March 2024 (the “Report”). Two points should be noted regarding this Report: first, the Report is not legally binding and, second, the Report does not necessarily present one reasonable or recommended view for each issue, but rather contains several possible views on various issues.

The Report discusses, among other things:

• the potential for copyright infringement during the course of developing generative AI;
• the potential for copyright infringement when using generative AI; and
• the potential for copyright protection for the outputs of generative AI.

Potential for copyright infringement during the course of developing generative AI

Overview

The Copyright Act does not provide a general defence such as the “fair use” defence under US copyright law on which alleged infringers may rely in the US. Instead, prior to amendment in 2018, the Copyright Act provided an exclusive list of defences covering specific situations in which copyrighted works could be used without authorisation from the copyright owner. The 2018 amendment of the Copyright Act was intended to expand the scope of defences and to introduce more generalised defences. The purpose of the 2018 amendment was to contribute to promoting innovation by enabling exploitation of copyrighted works without authorisation from the copyright owner, when such exploitation does not significantly affect the market of the copyrighted works. One of the generalised defences introduced by the 2018 amendment was the defence for an act with a non-enjoyment purpose, set forth in Article 30-4 of the Copyright Act. This defence was not as generalised as the “fair use” defence under US copyright law, but it was understood that Article 30-4 would allow for some flexibility in responding to new modes of use of works accompanying technological invention, compared to the more rigid defences available prior to the 2018 amendment.

Under Article 30-4 of the Copyright Act (as amended), a work may be exploited in any way and to the extent considered necessary, in cases where it is not the purpose of the exploiter to personally enjoy or cause another person to enjoy the thoughts or sentiments expressed in that work (the “Non-Enjoyment Purpose Requirement”). However, this exception does not apply if the relevant act would unreasonably prejudice the interests of the copyright owner in light of the nature or purpose of the work or the circumstances of the exploitation (the “Article 30-4 Proviso”).

The Report addresses the applicability of the foregoing defence with respect to infringement in the course of generative AI development by discussing the Non-Enjoyment Purpose Requirement and the Article 30-4 Proviso, and also discusses the parties against whom a copyright owner can exercise the copyright.

The Non-Enjoyment Purpose Requirement

Article 30-4 lists three examples in which it is not the purpose of the exploiter to personally enjoy or cause another person to enjoy the thoughts or sentiments expressed in a given work. The first example is the exploitation of a work for use in testing to develop, or to put into practical use, technology that is connected with audio recordings, video recordings or other exploitations of a copyrighted work. The second example is the exploitation of a copyrighted work for use in data analysis (meaning the extraction, comparison, classification, or other statistical analysis of the constituent language, sounds, images, or other elemental data from a large number of works or a large volume of other such data). The third example is the exploitation of copyrighted works – in the course of computer data processing or otherwise – in a way that does not involve what is expressed in the work being perceived by the human senses (for computer programming work, such exploitation excludes the execution of the works on a computer).

The Report states that the above-mentioned “data analysis” example includes that for the training of generative AI, and that the exploitation of a work for use in data analysis for training generative AI meets the Non-Enjoyment Purpose Requirement.

It should also be noted that the Non-Enjoyment Purpose Requirement is not satisfied (ie, the relevant exploitation can constitute copyright infringement) when the purpose of enjoyment and the purpose of non-enjoyment co-exist. The Report provides examples of cases in which the purposes of enjoyment and non-enjoyment are deemed to co-exist:

• a work is reproduced in order to perform additional training to intentionally output all or part of the creative expression of the copyrighted work contained in the training data as it is – eg, when an AI developer or AI service provider intentionally overfits a model; and
• a database where the content of works is converted into a vector for the purpose of outputting all or part of the creative expression of copyrighted works contained in an existing database or data posted on the internet.

In addition, the Report states that, even if the intent is not to output the creative expression of the copyrighted work, the purposes of enjoyment and non-enjoyment can nonetheless be found to co-exist. This is the case where:

• a work is reproduced for the purpose of additional training to generate outputs that are strongly influenced by the creative expression of the copyrighted work contained in the training data, by using a small amount of training data; and
• it is determined from the underlying facts of specific cases that the purpose is to generate outputs in which the creative expression of the copyrighted work contained in the training data can be directly perceived.

With respect to additional training for the purpose of generating outputs which imitate the “style” of the works of a specific creator, the Report states that “style” is just an idea and that even if AI output is in the same “style” as existing copyrighted works, that does not constitute copyright infringement, although the purposes of enjoyment and non-enjoyment can still be found to co-exist.

The Article 30-4 Proviso

Even when the Non-Enjoyment Purpose Requirement is satisfied, Article 30-4 does not apply where the interests of the copyright owner would be unreasonably prejudiced. Whether the interests of the copyright owner would be unreasonably prejudiced is determined by considering whether the subject act(s) conflict(s) with the market for the use of the copyrighted works of the copyright owner or whether the act(s) will hinder potential future sales channels for the copyrighted works. The Report discusses the following four situations:

• generation of many works which involve an “idea” similar to that of an existing copyrighted work;
• organisation of database works in a form that can be used for data analysis;
• taking technical measures to prevent the reproduction of works for training AI; and
• reproduction of infringing copies (such as pirated copies) for the purpose of training AI.

Remedy

If a copyright is infringed in the course of the development of generative AI, a copyright owner can seek injunctive relief, which includes measures necessary for the cessation or prevention of infringement, and compensation for damage. The Report discusses whether a copyright owner can seek, as part of injunctive relief, an order to remove their copyrighted works from a training data set to be used for future AI training, and to dispose of any models trained on the basis of such work.

The Report states that it is possible to obtain an order to remove the copyrighted works from training data sets. On the other hand, the Report states that an order to dispose of trained models is generally unavailable, except where it is highly likely that the trained model will generate outputs that are similar to the copyrighted works.

Possibility of copyright infringement in the use of generative AI

Overview

Copyright infringement is deemed to have occurred where there is both (i) similarity in the allegedly infringing work to an existing copyrighted work; and (ii) reliance on the existing copyrighted work (known as the “Reliance Requirement”).

Whether an AI output is “similar” to an existing copyrighted work in the above sense is determined in the same way as the similarity between a human-generated work and an existing copyrighted work would be.

The Reliance Requirement

The Report addresses three types of cases in which reliance may or may not be found, each of which concerns whether (i) the AI end user (the “AI User”) recognised the existing copyrighted work; and (ii) the existing copyrighted work was included in the training data, as follows:

• Where the AI User recognized the existing copyrighted work – the Report states that the Reliance Requirement is satisfied and it should be determined that the AI User is infringing the copyright.
• Where the AI User was unaware of the existing copyrighted work and the existing copyrighted work was included in the training data – the Report states that, in this scenario, the AI User will generally be considered to meet the Reliance Requirement, because the AI User had access to the existing copyrighted work. The Report also states, however, that if technical measures are taken which can ensure that the generative AI does not generate creative expressions of work used for AI training processes, then the Reliance Requirement may not be met, even if existing copyrighted work was included in the training data.
• Where the AI User did not recognise the existing copyrighted work and the existing copyrighted work was not included in the training data – the Report states that in this case, the Reliance Requirement is not met.

Remedies

A copyright owner may seek an injunction, including measures which are necessary for the cessation or prevention of infringement, as well as compensation for damage/damages. The Report states that a copyright owner can seek the following remedies as injunctive relief:

• against the AI User –
1. an order to refrain from generating infringing output data in the future and to cease using infringing output data already generated; and
2. an order to dispose of infringing output data already generated;
• against the AI developer – an order to remove the infringed copyrighted work from the data set that was used for the development of the AI (if it is likely that the data set will still be used for AI development); and
• against the AI service provider – an order to apply technical restrictions on the generative AI that generated the infringing output, such as –
1. measures to avoid generating output data in response to a specific prompt input; or
2. measures to avoid generating output data that is similar to the copyrighted works which were used for training AI (if it is likely that further copyright infringement will be caused by the generative AI that generated the infringing output).

In addition, the Report states that if generation and use of AI output data constitute copyright infringement, not only the AI User but also the AI developer or the AI service provider can be held liable for copyright infringement. The likelihood that the AI developer or the AI service provider will be held liable for copyright infringement will increase in the following situations:

• infringing output data is very frequently created when a specific generative AI is used; and/or
• the business operator does not take measures to avoid generation of output data that is similar to existing copyrighted works, even though it recognises there is a strong possibility that the generative AI will create such output data.

Governing law

With respect to the geographic coverage of the Copyright Act, the Report states that (i) as to damage claims, where the result of the copyright infringement is deemed to have occurred in Japan, Japanese copyright law will apply (as the law of the place where the result of the act in issue occurred); and (ii) as to claims seeking injunctive relief, Japanese copyright law, as the law of the country where protection is claimed, will govern where the act of exploitation is deemed to have taken place in Japan.

The Report states that the following factors will increase the possibility that Japanese copyright law will apply:

• in the course of the development/training of generative AI, the program collecting data for training AI ran on a server located in Japan, and existing copyrighted works were reproduced in connection therewith;
• where output data, which included existing copyrighted works, was produced by generative AI running on a server located in Japan; or
• where output data which included copyrighted works was produced by generative AI and publicly transmitted to users in Japan by an AI service provider.

Interim Report of the Study Group on Intellectual Property Rights in the AI Era (May 2024)

In October 2023, the Study Group on Intellectual Property Rights in the AI Era (the “Study Group”) was convened. The Study Group organised discussions regarding the relationship between AI and intellectual property rights, based on considerations of this in relevant ministries and agencies, to consider necessary measures, etc. The Study Group mainly discussed and considered the following two issues: (i) measures with respect to concerns and risks regarding generative AI and intellectual property; and (ii) the protection of inventions based on developments in AI technology.

In May 2024, following discussion of the foregoing issues, the Interim Report of the Study Group on Intellectual Property Rights in the AI Era (the “Interim Report”) was published. The Interim Report is not legally binding or a definitive legal assessment. The Interim Report does, however, set out the results of these discussions based on considerations in the relevant ministries and agencies, concerning various issues related to generative AI and IP rights. Therefore, the Interim Report is a valuable resource for parties engaged in businesses related to generative AI in Japan.

The Interim Report states that: (i) the Intellectual Property Act; (ii) technology; and (iii) the contract are all needed to deal with concerns and risks regarding generative AI and intellectual property, as each measure cannot cover all the concerns and risks by itself.

Intellectual Property Act

The Interim Report introduces issues in respect of the Copyright Act in the development stage and in the use stage of generative AI, by referring to the Report (see “Report on AI and Copyright Issues (15 March 2024)” above). The interim Report also discusses the potential for design infringement, trade mark infringement and unfair competition in the two stages of generative AI. As to trade mark infringement, while Japan’s Trademark Act sets no regulation applicable to acts during the course of developing generative AI, trade mark infringement is fundamentally deemed to have occurred when using generative AI where there is similarity between: (i) the alleged infringing mark that contains AI output; and (ii) an existing trade mark. However, when determining whether trade mark infringement has occurred in the development of a mark, it makes no difference if the output was generated by generative AI or not. On the contrary, a mark can be protected by the Trademark Act regardless of whether it was created by a person or generative AI.

Technology

The Interim Report points out that the development of AI technologies is ongoing, and therefore, to assess and avoid risks in terms of generative AI and intellectual property infringement, two points should be considered: (i) what specific technological measures can be employed to cope with issues of generative AI; and (ii) how people can use technological measures in order to ensure and promote the use of these technological measures. The Interim Report refers to five examples of technological measures:

Firstly, a system enabling users to recognise when an output was generated by AI (digital watermark, etc). With respect to such a system, the Interim Report points out that the existence or non-existence of an indication of the generation of AI output does not always lead to the existence or non-existence of the copyrightability of the AI output. Further discussion is expected regarding who attaches the indication, and the scope of AI output to which the indication should be attached.

Secondly, the use of filtering, which is technology to determine whether AI output is similar to other content or to suppress AI input/output of data/content that may infringe intellectual property rights. This is a useful measure to avoid copyright infringement, to some extent. However, the Interim Report notes that the similarity element under the Copyright Act should be determined from the perspective of whether both works have creative expression. Therefore, accurately judging the similarity element may be difficult if using filtering which checks mere external similarity.

Thirdly, the Interim Report introduces technology to refuse collection by automatic collection programs called “crawlers”. For example, there is a system, called “robots.txt” that gives instructions about whether crawlers can access specific content (however, “robots.txt” is ineffectual when the crawler ignores it). Further, restricting crawlers’ access by means of ID and password is also effective.

Fourthly, the introduction of technology to prevent learning by applying special image processing. This technology is useful from the point of view that copyright owners can directly control situations in which their works are used for AI learning, since it is impossible to generate new learning images with similar styles using this technology. However, the Interim Report notes that the use of such technology for the purpose of interference in the business of AI developers and providers may constitute the crime of obstruction of business by damaging a computer.

Fifthly, there is technology related to individual tracking and exclusion from learning source content and assimilating data. However, under the current technology, it is difficult for users of trained models to check the original data used to train the AI. Further discussion based on future developments of such technology is needed regarding whether it is appropriate for AI developers to be obliged to exclude certain data at a copyright owner’s request.

Contract

The Interim Report notes that, to promote AI technologies and preserve IP rights, it is necessary to put in place measures by which creators (copyright owners) will be sufficiently compensated for the use of generative AI, and that provide incentives for new creative activities. From this viewpoint, the Interim Report provides three examples of how creators can be compensated:

• creators themselves provide training data for additional learning (fine-tuning) for profit;
• creators themselves develop and provide generative AI; and
• creators themselves use generative AI in their creative activities.

To facilitate such measures for compensation, it would be beneficial for the parties involved to confirm the legal rules that guarantee the above examples of contractual measures, and to employ technical measures that, in turn, can guarantee contractual measures.

Other Developments in 2024

On 19 April, the Ministry of Economy, Trade and Industry (METI) released the AI Guidelines for Business Ver 1.0, which set out the AI governance regime in Japan (find the English version here). The guidelines are intended for all those who use AI in various activities and they set out the collaborative efforts of each entity, such as AI developers, AI providers and AI users. On 5 July, METI released the Guidelines for Utilization of Generative AI for Content Creation, which also set out points to keep in mind and measures to be taken for each setting of use of generative AI in each industry. On 31 July, the Cultural Agency released the Checklist & Guidance on AI and Copyright, detailing checklists for AI developers, AI providers, AI users, and non-professional (general) users, and how to deal with copyright infringement. Both the Guidelines for Utilization of Generative AI for Content Creation and the Checklist & Guidance on AI and Copyright are based on the Report on AI and Copyright Issues (see above).

Conclusion

Although the above guidelines are not legally binding, checking and following them is of significant value, since there are no specific laws and regulations regarding generative AI.

With regard to developments in respect of precedents, there have been no court decisions on AI or on copyright or trade mark rights, and no specific cases of disputes have been reported (with respect to patent rights, Tokyo District Court ruled on 16 May 2024 that “inventor” as defined in the Patent Act of Japan is limited to natural persons and does not include generative AI). Future developments regarding generative AI and intellectual property rights in Japan will be monitored closely.