Artificial Intelligence (“AI”) is readily becoming a tool for the immediate creation of written works, images and designs (and much more) with very little effort or human input required. AI learns from the information stored in its system (including images, designs, art and literary works) and can generate responses to prompts from users based on that information. However, as AI becomes more accessible and usage increases in New Zealand (particularly in commercial applications), intellectual property ownership becomes an important matter.
The use of AI poses several questions about copyright ownership, including potential copyright infringement by AI generated content, and ownership of the AI generated content.
Copyright Infringement
Data used as inputs or training material for AI
In 2023 Getty Images brought a copyright infringement lawsuit in the United Kingdom against the AI company, Stability AI Inc. Getty Images claims that Stability AI Inc copied millions of its images from the internet to ‘train’ its AI software to produce images from text inputs by users. Stability AI software supposedly generated images with the “Getty Images” watermark, which Getty said proved that Stability AI used Getty Images to train its AI. The lawsuit has yet to be determined but raises the question of whether using unlicensed data and information to train AI software is an infringement of copyright.
In New Zealand, copyright is a property right that subsists in original works, including literary, musical, and artistic works.[1]
Copyright infringement occurs when a person uses, copies or communicates the work of another person without their permission.[2]
There are a number of exceptions to copyright infringement, including news reporting, research or private study, or education.[3]
Copying someone else’s copyrighted works in order to ‘train’ AI does not appear to fall under any of the exceptions and may be a breach of New Zealand copyright laws. This is different from the United States, which has much broader “fair use” exceptions to copyright protection and allows copying of works where it is not for a commercial purpose and it is used to create something new (or for transformative purposes).[4]
While there is no case law on copyright infringement by AI in New Zealand, it is interesting that Stuff.co.nz has joined many other online news organisations in preventing AI software from scraping data from its website to train AI.[5] As AI becomes more prominent, companies appear to be taking copyright and intellectual property issues into their own hands.
AI Generated Content
AI takes the information that it has used in its ‘training’ and generates an output (whether that is an image, design, art or written text) from what it has ‘learnt’. If AI has been trained using information that is protected by copyright, then the AI generated content may include substantial parts of work protected by copyright.
In the near future, information used to train AI may be licensed and there may be no risk of AI generated content infringing on copyright. However, at this stage it appears some AI companies are scraping data from the internet (irrespective of copyright and other intellectual property rights) to train their AI. Therefore, users of AI should be aware that AI content may be a derivative of a copyright work and be in breach of copyright laws.
Notably, Microsoft appears to be introducing safeguards and guardrails to prevent its AI software from generating copyright infringing content. Microsoft has also announced that it will defend its customers who are subject to copyright infringement proceedings as a result of using paid Microsoft Copilot services (type of AI). [6]
Implications of generating AI content
New Zealand’s copyright law on AI generated content is different to the United States.
Recently, a Federal Court Judge held that AI generated content was not registerable for copyright. This decision was based on the underlying principle that copyright must have some human involvement.[7] Under New Zealand law, the Copyright Act 1994 specifies that computer-generated works can be protected by copyright and copyright will vest in “the person by whom the arrangements necessary for the creation of the work are undertaken”.[8] This means the person who inputs prompts into the AI software could be the copyright owner of the AI generated content. Users of AI will still need to establish that the work is original and that they were the ‘author’ if they wish to assert copyright in the generated work.[9] It may therefore be important to record the contribution that users had in creation of a work if AI was used.
AI Software Terms and Conditions
Ownership of AI generated content may also be subject to the specific terms of use for the AI software.
For example, the Stability AI Inc terms specify that any content created by its AI software is owned by Stability AI Inc and not the user.[10] The Stability terms also specify that users will grant a perpetual, worldwide, non-exclusive, sub-licensable, royalty free, irrevocable copyright licence to Stability AI Inc for any content collected (i.e. prompts used to generate the content, whether a word or image prompt). Consequently, this means that Stability AI Inc may be able to distribute or use content it collects from the user to create and supply derivative works to its other customers.
In comparison, the OpenAI (ChatGPT) terms assigns all right, title and interest in the AI content to the user.[11] However, the user is responsible for the AI content and for making sure that it does not violate any laws or the OpenAI terms. The OpenAI terms disclaim all warranties, including warranties as to merchantability, fitness for purpose and non-infringement. Therefore, users should keep in mind that by using ChatGPT they agree that use of AI content is at their own risk. Additionally, the OpenAI terms provide that businesses or organisations using the software will indemnify OpenAI for any costs, losses, liabilities and expenses from any third-party claims arising out of the AI generated content or use of OpenAI’s services.
OpenAI may also use the AI content to train its models unless this is opted out of in the ChatGPT settings.
Generating AI content may have varying implications on users depending on what AI software they use. Therefore, it will be important for users to read and consider the terms of use for any AI software that they intend to use.
Summary
Content creators should be mindful when using AI to create works. AI generated content may be a derivative of a copyrighted work, depending on the source and information used to train the AI software.
Content creators should also consider whether information or works can be easily scraped from their websites by AI companies and whether copies or derivatives of their works are already being generated by AI.
Anyone using AI should be conscious of the terms and conditions that each AI software implements and who owns the content generated by the AI software.
The AI and intellectual property landscape is quickly unfolding and will likely become a normal consideration for all content creators in the near future.
Disclaimer
This article is provided to assist clients to identify legal issues on which they should seek legal advice, and by its nature cannot be comprehensive and cannot be relied on as advice. Please consult the professional staff of Lane Neave for advice specific to your situation.
[1] Copyright Act 1994, s14.
[2] At ss 30, 31 and 33.
[3] At ss 41 – 43.
[4] The Copyright Act 17 USC, § 107.
[5] See https://www.stuff.co.nz/business/132917438/stuff-joins-media-firms-blocking-chatgpt-from-learning-from-their-content.
[6] https://www.microsoft.com/en-us/licensing/news/microsoft-copilot-copyright-commitment.
[7] Thaler v Perlmutter, D.D.C., No. 1:22-cv-01564.
[8] Copyright Act, at s5.
[9] At ss14 and 18.
[10] https://stability.ai/discord-tos.
[11] https://openai.com/policies/terms-of-use.
