Generative AI usage guidelines
1. Purpose of these Guidelines
This guideline explains the points that OpenFashion Co., Ltd. should be aware of when using generative AI such as MaisonAI, ChatGPT, and Midjourney in its business operations.
While generative AI can be useful for improving business efficiency and generating new ideas, depending on the content of the data entered and how the generated results are used, it may violate laws and regulations or infringe on the rights of others. Please read these guidelines carefully and use generative AI effectively.
2. Generative AI covered by these guidelines
The generative AI covered by this guideline is MaisonAI, ChatGPT, and Midjourney. If you wish to use a generative AI other than these, please contact the Corporate Department.
3. Prohibited uses of generative AI
At OpenFashion, we prohibit the use of generative AI for the following purposes and business purposes.
(1) Analysis and use of personal information of customers obtained through our own services and client business
(2) Analysis and use of confidential information obtained in the course of client work
4. Structure of these Guidelines
The basic structure of all generative AI services is that "the user inputs some kind of data, some kind of processing is carried out (storage, analysis, generation, learning, re-provision, etc.), and the result (product) is obtained."
Therefore, the guidelines consist of two parts:
- Points to note when entering data
- Precautions to take when using the product
5. Points to note when entering data
The data that can be input (sent) to generative AI can be of a wide variety, but from the perspective of the need to process intellectual property rights and comply with laws and regulations, particular care must be taken when inputting the following types of data:
(1) Data copyrighted by a third party (such as text created by another person)
In principle, simply inputting someone else's copyrighted work into a generative AI does not constitute copyright infringement.
However, if the purpose is to generate an AI product that is identical or similar to the input work of another person, the act of input itself may constitute copyright infringement.
In addition, if the generated data is identical or similar to the input data or existing data (copyrighted work), please note that using the product may infringe the copyright of the copyrighted work. For details, please refer to "6 (2) Using the product may infringe someone's existing rights."
In addition, in principle, creating original models through fine tuning or using the works of others for so-called prompt engineering is not considered to constitute copyright infringement.
(2) Registered trademarks and designs (logos and designs)
Inputting logos, designs, etc. that are registered as trademarks or designs into a generation AI does not constitute trademark or design right infringement.
However, it should be noted that this only applies to the act of inputting, just as with copyrighted works. Commercial use of a trademark or design that is identical or similar to another person's registered trademark or design, whether created intentionally or accidentally, constitutes trademark or design infringement.
In other words, there is little need to research registered trademarks or designs when inputting a logo or design into a generative AI, but research is necessary if the generated result is to be used.
(3) Photographs or names of celebrities
Inputting a celebrity's photograph or name into a generation AI does not constitute an infringement of the celebrity's right of publicity.
However, please note that commercial use of the names, portraits, etc. of celebrities generated using generative AI constitutes a violation of their publicity rights.
(4) Personal information
In ChatGPT, the data entered is used to train OpenAI's model, so if you enter personal information (customer name, address, etc.) in ChatGPT, you must obtain the consent of the person identified by the personal information. In principle, entering personal information in other tools is also prohibited. Please contact the Corporate Department for details.
(5) Confidential information disclosed by another company under a confidentiality obligation
Inputting confidential information obtained through a non-disclosure agreement (NDA) or other agreement with another company into a generation AI provided by an external company may violate the NDA, as it would amount to "disclosure" of the confidential information to a "third party" - the generation AI provider.
Therefore, please do not enter any such confidential information.
(6) Confidential information of the organization
Inputting your company's confidential information (know-how, etc.) into a generation AI does not violate any laws or regulations. However, depending on the processing content of the generation AI and the contents of the terms and conditions, there is a risk that the confidential information may no longer be protected by law or that a patent application may no longer be possible, so please do not input it.
6. Precautions to be taken when using the product
(1) There is a possibility that the content of the product may contain falsehoods.
The principle of large-scale language models (LLM) is to create plausible sentences by outputting "the word that is most likely to be used next to a certain word." The contents of the sentences may contain lies.
Please be aware of these limitations of generative AI, and do not blindly trust the contents of its products; always check the reasons and supporting evidence for yourself.
(2) Using the work may infringe someone's existing rights.
① Copyright infringement
If the output generated by generative AI is identical or similar to an existing copyrighted work, using that output (such as copying or distributing it) may constitute copyright infringement.
Therefore, please observe the following points:
- Please do not use specialized AI that has been trained only on the works of a specific author or writer.
- Please do not enter the name of an existing work, author, or piece at the prompt.
In particular, when you "use" the result (distribute, publish, etc.), be sure to investigate whether the result is similar to existing copyrighted works and consider whether the use of the result falls under any restrictive provisions (Article 30 Paragraph 1 or 30-3 of the Copyright Act, etc.).
② Infringement of trademark and design rights
Using images generated by image generation AI or catchy slogans generated by text generation AI in product logos or advertising may infringe on the registered trademarks or registered design rights held by others. Therefore, in addition to investigating whether the resulting work is similar to existing copyrighted works, you should also conduct a search for registered trademarks and designs.
3) False personal information, defamation, etc.
ChatGPT and other services are known to have the potential to generate false information about individuals. Creating and using/providing false personal information may be a violation of the Personal Information Protection Act (Articles 19 and 20) or defamation or slander, so please refrain from such actions.
(3) There is a possibility that no copyright will be generated for the resulting work
If a creation did not have copyright, then that creation would essentially be open to any third party to copy, which would be a major problem for individuals and organizations who want to protect the rights to their own creative works.
On this point, the conclusion depends on whether or not there is a human "creative contribution" to the creative activity using generative AI, so we recommend that you avoid using the generated results as is and instead add to and revise them as much as possible.
(4) The product may not be commercially usable.
When using products generated by generative AI for business purposes, the question arises as to whether the products can be used commercially.
This issue will vary depending on the terms of use and contract conditions of the generative AI you use, so please use commercially available tools.
(5) Be aware of policy restrictions on generative AI
In addition to the risks (mainly legal restrictions) that we have explained so far, generative AI may also impose its own restrictions based on its service policies.
When using ChatGPT, please note the following:
The Usage Policies (https://openai.com/policies/usage-policies) specify specific prohibitions, such as "Adult content, adult industries, and dating apps" and "Engaging in the unauthorized practice of law, or offering tailored legal advice without a qualified person reviewing the information."
The policy also states that businesses that create and provide content for consumers, such as those in the healthcare, finance and legal industries, news generation and news summarization, must provide disclaimers to users informing them of the use of AI and its potential limitations.
In addition, the relevant policies stipulate that when publishing content generated using OpenAI's services, such as ChatGPT, it is necessary to clearly indicate that it is an AI-based product.
When using other tools, please be sure to understand and comply with their respective policies.