Can AI-generated art be used for commercial purposes? This is a question that is increasingly being asked as the use of artificial intelligence (AI) in art becomes more prevalent. On the one hand, AI-generated art offers new possibilities for users, including the right to use their generations for commercial purposes. On the other hand, concerns have been raised around who owns the images and if they might infringe on existing copyrighted works.
As of now there is no clear law on the copyright of AI images. According to the AI developers, all are free to use the art generated by their AI image generators for commercial use but experts say if the model is trained using a copyrighted image, then it cannot be used.
How AI models work?
One example of AI-generated art is text-to-image generators such as DALL-E, Stable Diffusion, and Midjourney. These AI models can create images based on a text prompt by interpreting the text and combining different concepts, attributes, and styles. To make this possible, the models are trained using a massive number of images. For example, OpenAI’s DALL-E 2 is trained on around 650 million images from a mix of publicly available sources and “sources we licensed.” However, the company has not made the dataset public, leading to concerns that copyrighted material could be within it.
The use of AI-generated art for commercial purposes raises questions about copyright ownership. Copyright protects original works of authorship, such as literary, musical, or artistic works. When a person creates a work of art, they own the copyright to that work. However, when an AI generates a work of art, it is unclear who owns the copyright. This is because copyright law was created for human-created works, not machine-created works.
What experts are saying?
IP law expert Bradley J Hulbert recently told TechCrunch that AI-generated images could cause various problems from a copyright perspective. He said that artwork that bears a resemblance to a “protected work” such as a Disney character or logo needs to be “transformative” to be legally protected. If a piece of work qualifies as fair use under a legal defense such as this, then it would not be considered a copyright infringement. However, the issue around fair use protection becomes confusing when AI is involved. There is no direct legal precedent in the US that upholds publicly available training data as fair use.
The datasets that these AI models are trained on are also crucial. The models could potentially create new images that “mirror” an original, leading to risks of copyright infringement. It is important that the datasets used to train these models are scrutinized and do not contain copyrighted material.
There are some efforts being made to address these concerns. For example, Stable Diffusion is being released under a Creative ML OpenRAIL-M license, which allows for commercial and non-commercial usage. The license is focused on ethical and legal use of the model and must accompany any distribution of the model. It must also be made available to end users of the model in any service on it. The company has also developed an AI-based Safety Classifier included by default in the overall software package. This understands concepts and other factors in generations to remove outputs that may not be desired by the model user.
Conclusion
The use of AI-generated art for commercial purposes raises important questions about copyright ownership and fair use. It is important that the datasets used to train these models are scrutinized and do not contain copyrighted material. Companies should also take steps to ensure that the models are released under ethical and legal licenses and that end users are aware of their responsibilities when using the models. While there are challenges to be addressed, the use of AI in art offers exciting new possibilities for creators and users alike.
If planning to use an AI model for commercial purpose, it’s better to train the model with open-source images for a specific purpose and use the trained model. Thus, making sure that the model doesn’t generate clone of copyrighted image.