There are two recent law cases about AI image generators that I found. One is a class-action lawsuit filed by a group of artists against several AI companies, including Stable Diffusion, Artbreeder, DeepArt, and RunwayML. The artists claim that these companies have used their artworks to train their AI models without their permission and that the AI-generated images infringe on their copyrights.

The other case is a lawsuit filed by Getty Images against Stable Diffusion, the creator of an open-source AI art generator. Getty Images alleges that Stable Diffusion has scraped millions of images from its website and used them to train its AI model, which violates its terms of service and infringes its rights.

Both cases are pending in the US federal court in San Francisco, and they raise some interesting and complex legal questions about the use and ownership of AI-generated images.

Lawyer Explains Stable Diffusion Lawsuit

What are the legal questions involved in these cases?

Some of the legal questions involved in these cases are:

• Who owns the copyright of AI-generated images?
• Are the AI-generated images original or derivative works?
• Do the AI-generated images infringe the rights of the original artists or image owners?
• How can the original artists or image owners prove their ownership and damages?
• What are the ethical and social implications of using AI-generated images?

These questions are not new, but they are becoming more relevant and challenging as the technology of text-to-image AI generators advances and becomes more accessible. Text-to-image AI generators are online tools that create images based on simple text prompts, using artificial intelligence to learn from a training set of images. The images they generate can be quite impressive, but also potentially problematic.

What are the legal implications of using AI-generated images?

Some of the legal implications of using AI-generated images are:

• You may not have full rights to use the AI-generated images, especially if they are based on images that belong to someone else. This can result in disqualification, infringement claims, or legal actions from the original image owners.
• You may not be able to claim ownership or protection for the AI-generated images, especially if they are not original or creative enough. This can result in a loss of revenue, reputation, or competitive advantage.
• You may not be compliant with data protection laws, especially if the AI-generated images involve personal data or sensitive information. This can result in fines, sanctions, or lawsuits from the data subjects or regulators.

Therefore, when using AI-generated images, it is important to fully understand the ownership rights and any restrictions that may apply and to obtain the necessary permissions or licenses from the relevant parties.

How can I obtain permissions or licenses for AI-generated images?

The answer to this question may depend on the source and nature of the AI-generated images, as well as the intended use and purpose of the images. However, some general steps that you can take to obtain permissions or licenses for AI-generated images are:

• Check the terms and conditions of the AI image generator that you are using, and see if they grant you any rights or licenses to use the images commercially or non-commercially. For example, OpenAI announced that users have full rights to commercialize the images they create with its DALL.E 2 tool.
• Check the originality and creativity of the AI-generated images, and see if they qualify for any protection or registration under the relevant laws and jurisdictions. For example, the U.S. Copyright Office ruled that AI-generated images that lack human authorship cannot be copyrighted.
• Check the ownership and rights of the images that were used to train the AI model, and see if you need to obtain any permissions or licenses from the original image owners or licensors. For example, Getty Images sued Stable Diffusion for scraping millions of images from its website and using them to train its AI model without permission or compensation.
• Check the content and context of the AI-generated images, and see if they comply with any ethical, social, or legal standards or regulations. For example, you may need to protect personal data or sensitive information in the AI-generated images, or avoid using AI-generated images that are misleading, harmful, or offensive.

These steps are not exhaustive, and you may need to consult a legal expert or a professional service for more specific guidance and advice.

How can I check the originality or creativity of AI-generated images?

Checking the originality or creativity of AI-generated images can be a challenging and subjective task, as there is no clear or universal definition or standard for these concepts. However, some possible ways that you can check the originality or creativity of AI-generated images are:

• Compare the AI-generated images with the images that were used to train the AI model, and see if they are substantially similar or different. If the AI-generated images are too similar to the training images, they may not be original or creative enough. If the AI-generated images are too different from the training images, they may not be relevant or realistic enough.
• Compare the AI-generated images with the images that are available online or offline, and see if they are unique or common. If the AI-generated images are too common or generic, they may not be original or creative enough. If the AI-generated images are too unique or novel, they may not be appealing or understandable enough.
• Evaluate the AI-generated images based on your own criteria and preferences, and see if they meet your expectations or goals. If the AI-generated images are too simple or boring, they may not be original or creative enough. If the AI-generated images are too complex or confusing, they may not be effective or efficient enough.

These ways are not definitive, and you may need to use a combination of them or consult an expert or a professional service for more accurate and reliable results.

How can I protect personal data or sensitive information in AI-generated images?

Protecting personal data or sensitive information in AI-generated images can be a crucial and complex task, as there are various risks and challenges involved, such as data breaches, identity theft, privacy violations, ethical dilemmas, and legal compliance issues. However, some possible ways that you can protect personal data or sensitive information in AI-generated images are:

• Use data anonymization tools and techniques to remove or obscure any identifying or sensitive information from the images, such as names, faces, locations, or biometrics. This can help reduce the risk of exposing or misusing personal data or sensitive information in AI-generated images.
• Use data classification methods and policies to determine the level of sensitivity and business impact of the images, and apply the appropriate security measures and controls accordingly, such as encryption, access control, or auditing. This can help manage and monitor personal data or sensitive information in AI-generated images.
• Use data protection laws and regulations to guide and govern the collection, processing, and sharing of the images, and ensure that you have the necessary consent, purpose, and accountability for using the images. This can help comply with the legal and ethical standards and obligations for personal data or sensitive information in AI-generated images.

These ways are not exhaustive, and you may need to use a combination of them or consult a legal expert or a professional service for more specific guidance and advice.

What are some examples of AI-generated images that are original and creative?

There are many examples of AI-generated images that are original and creative, as AI models and tools can produce images based on various inputs, such as text, sketches, or other images. Some examples of AI-generated images that are original and creative are:

• DALL.E 2 is an AI model developed by OpenAI that can generate images based on text prompts, such as “a photo of a quaint flower shop storefront with a pastel green and clean white facade and open door and big window” or "a Shiba Inu wearing a beret and black turtleneck". The images generated by DALL.E 2 are original and creative because they can capture the details and nuances of the text prompts, and also produce variations and combinations that are unexpected and imaginative.
• Midjourney is an AI tool that can generate realistic and high-quality images of people, animals, landscapes, and objects, based on sketches or keywords, such as “woman with curly hair and glasses” or "mountain lake at sunset"³. The images generated by Midjourney are original and creative because they can fill in the gaps and enhance the features of the sketches or keywords, and also create images that are diverse and appealing.
• Hypotenuse AI is an AI platform that can generate stunning and unique images for e-commerce, social media, and marketing, based on product descriptions or brand names, such as “a pair of black leather boots with buckles and studs” or "a logo for a coffee shop called Bean There". The images generated by Hypotenuse AI are original and creative because they can match the style and tone of the product descriptions or brand names, and also create images that are relevant and attractive.

These examples are not exhaustive, and there are many other AI models and tools that can generate images that are original and creative, depending on the input and output desired.

What are some benefits of using AI-generated images?

There are many benefits of using AI-generated images, as AI models and tools can create images that are faster, more accurate, and more diverse than human-made images. Some benefits of using AI-generated images are:

• Increased efficiency: AI-generated images can be created much faster than manually created images, saving time and effort when compared to creating each image from scratch. This can help with tasks like generating large numbers of images for e-commerce, social media, or marketing campaigns.
• Enhanced accuracy: AI-generated images can accurately replicate images and make them more realistic, which can help with tasks like training machine learning models, testing software, or simulating scenarios. AI-generated images can also avoid human errors or biases that may affect the quality or validity of the images.
• Expanded creativity: AI-generated images can create images that are more diverse and novel than human-made images, which can help with tasks like exploring new ideas, expressing emotions, or inspiring audiences. AI-generated images can also combine or transform images in ways that are unexpected and imaginative.

These benefits are not exhaustive, and there may be other benefits of using AI-generated images, depending on the purpose and context of the images.

What are some risks of using AI-generated images?

There are also some risks in using AI-generated images, as AI models and tools can create images that are infringing, misleading, or harmful to human rights and values. Some risks of using AI-generated images are:

• Infringement of intellectual property rights: AI-generated images can infringe the intellectual property rights of the original creators or owners of the images that are used as inputs or sources for the AI models and tools. AI-generated images can also infringe the intellectual property rights of existing images that are similar or identical to the AI-generated images, even if the AI models and tools did not have access to them. This can lead to legal disputes and liabilities for the users of the AI-generated images.
• Misrepresentation of reality or facts: AI-generated images can misrepresent reality or facts by creating images that are fake, altered, or manipulated, which can deceive or confuse the viewers or users of the images. AI-generated images can also misrepresent reality or facts by creating images that are biased, inaccurate, or incomplete, which can distort or omit the truth or context of the images. This can lead to ethical issues and social harm for the users and the society of the AI-generated images.
• Violation of human dignity or privacy: AI-generated images can violate human dignity or privacy by creating images that are offensive, abusive, or invasive, which can harm or exploit the subjects or targets of the images. AI-generated images can also violate human dignity or privacy by creating images that are impersonating, identifying, or tracking, which can infringe or endanger the rights or interests of the individuals or groups of the images. This can lead to moral dilemmas and human rights violations for the users and the subjects of the AI-generated images.

These risks are not exhaustive, and there may be other risks of using AI-generated images, depending on the quality and purpose of the images.

What are some ethical issues of using AI-generated images?

Some ethical issues of using AI-generated images are:

• Authorship and ownership: AI-generated images raise questions about who is the author and owner of the images, and who has the rights and responsibilities over them. Is it the AI model or tool that created the images, the user who used the AI model or tool, the original creator or owner of the images that were used as inputs or sources, or someone else? How can the authorship and ownership of AI-generated images be determined, recognized, and protected?
• Credit and compensation: AI-generated images raise questions about how to credit and compensate the artists and creators who contributed to the images, either directly or indirectly. How can the AI model or tool acknowledge and respect the originality and creativity of the artists and creators whose images were used as inputs or sources? How can the user who used the AI model or tool share and reward the benefits and profits of the images with the artists and creators? How can the artists and creators protect and claim their rights and interests over the images?
• Artistic freedom and evolution: AI-generated images raise questions about how to balance artistic freedom and evolution with ethical values and norms. How can the AI model or tool ensure that the images are not offensive, abusive, or harmful to the subjects or targets of the images, or to the society and culture of the images? How can the user who used the AI model or tool respect and appreciate the diversity and novelty of the images, without compromising the quality and validity of the images? How can artists and creators embrace and adapt to the changes and challenges that AI brings to the art world?

These ethical issues are not exhaustive, and there may be other ethical issues of using AI-generated images, depending on the context and impact of the images.

Conclusion

AI generators are innovative and creative tools that can produce original and diverse images based on text prompts. However, they also pose significant challenges and risks to the intellectual property rights of the artists and creators whose images are used as inputs or sources, or whose images are similar or identical to the AI-generated images. The current legal frameworks and systems are not well-equipped to deal with the complex and dynamic issues of authorship, ownership, infringement, and liability that arise from the use of AI generators. Therefore, there is a need for more research, dialogue, and collaboration among the stakeholders involved, such as the AI developers, users, artists, creators, and policymakers, to find fair and effective solutions that can balance the interests and values of all parties, and foster a responsible and ethical use of AI generators.