Intellectual Property Rights in the Age of AI
Artificial intelligence (AI) has ushered in a new era of innovation, fundamentally transforming industries, processes, and the way we create. From AI-generated artwork to sophisticated software algorithms that can write music, code, create art, summarize information, and even draft legal documents, the capabilities of AI are expanding rapidly. The evolution of AI challenges our traditional notions of intellectual property (IP) rights, raising complex legal and ethical questions about ownership, authorship, and copyright in the digital age.
In this article, we will examine the challenges technologists and creators that use AI-based tools face obtaining ownership of their creations and how existing law addresses these issues. The human/machine interaction creates ambiguities in the law that will be resolved over time. However, concepts of fair use, transformation, inventorship/authorship, conception, and exhaustion answer many of the questions that have arisen from broad access to AI-systems.
The Challenge of AI Creations
The primary challenge facing AI users is: Who owns the output? Should AI-generated creations be protected under current patent and copyright laws or should AI-generated creations be forfeited to the public domain?
To date, courts and administrative agencies have steadfastly refused to grant IP rights in any AI-generated creations, relying on the notion that copyrights and patents must be the result of human activity. For example, copyright protection is traditionally granted to works that are original and created by human authors. The U.S. Patent law has been interpreted as including “anything under the sun that is made by man." The implications of this interpretation have driven decisions relating to isolated genes and computer software. The requirement for human authorship and being “made by man” leaves AI-generated works in a legal gray area, as these creations may not meet the traditional criteria for copyright protection.
Who Owns the Creations?
The nature of AI-generated content raises questions of authorship/inventorship which leads to questions of ownership. For example, if an AI creates a piece of art or writes a novel, who owns the copyright to that work? Is it the developer who created the AI, the user who initiated the creative process, or does the AI itself the author?
This question may be further complicated by various ownership interests in the data used to train the AI model. Although the concepts of fair use and transformative use of copyrighted materials that have been developed over the last 200 years provide a framework for answering these questions, data aggregators and owners of copyrighted works have questioned how/whether these concepts apply to AI training data.
The real question is: How will questions of authorship/inventorship and questions of ownership of training data impact innovation? Some argue for the extension of copyright protection to AI-generated works, to incentivize innovation and ensure that creators that use AI-based tools can benefit from their creations. Others caution against overly broad copyright protections that could stifle creativity and hinder advances in science, literature, and art that AI-based tools could provide.
Does copyright and patent law need to evolve to address the realities of a world with pervasive computational creative tools or does the law developed over the last 150 years encompass AI-generated works? Time will tell.
Potential Legal Frameworks and Solutions
The legal framework underlying modern IP law is sufficiently robust to accommodate AI-generated creations if the question of ownership can be resolved. This, unfortunately, is no small feat. Several approaches have been brought forward to help bridge the gap:
Attribution and Ownership Models: Creating clear guidelines for attributing AI-generated works and establishing authorship/inventorship will clarify the legal status of such creations. This will likely involve recognizing the human contributors behind the AI, or developing new categories of IP rights tailored to AI innovations.
For example, recognizing that the human querying the AI system, i.e. the “Prompt Engineer,” contributes the “conception” of the creation and controls the AI-based tools activity in the same way an inventor controls the activity of laboratory staff provides clear evidence that the human and not the AI system is the “inventor” of the creation under U.S. patent law.
Flexible Standards: Introducing more flexible standards that can accommodate the unique aspects of AI creations without undermining the value of human authorship/inventorship is another approach. This might include revising the criteria for copyrightability and patentability or introducing a tiered system of protection based on the level of human involvement.
While “conception” provides a legal framework for inventorship of patents, the authors of copyrighted works rarely come under scrutiny. Attributing authorship may therefore require a new framework that recognizes the contributions of both the human prompt engineer, the designer of the AI-system, and potentially the owner of training data.
Open Licensing and Collaboration: Encouraging the use of open licenses and collaborative frameworks can promote the sharing and collective improvement of AI technologies while still protecting the rights of creators. This approach can help balance the need for innovation with the importance of access and dissemination.
Software often includes open licenses that allow developers to use the code free of charge. However, the potential value of AI-based systems, which can be extremely high, provides a big incentive for businesses to keep AI-systems proprietary.
Open Access Questions: More important question is: How open access to copyrighted materials on the internet impacts the owners rights? The concept of patent or copyright “exhaustion” gives the public the rights to the invention or work, eroding the owners' enforcement rights. For example, copyright holders who make news articles available to the general public free of charge have limited enforcement rights for training data, e.g. they may not be able to stop AI developers from using the article as training data, and they may not be able to extract royalties for outputs relating the article because copyright exhaustion limits these rights.
Conclusion
As we venture further into the age of AI, it is crucial to understand the existing legal framework as it applies to creators that use AI-based tools to encourage innovation while protecting the rights of the creators. By applying existing laws to the realities of the digital age, we can ensure that the future of creation remains bright, diverse, and open to all.
While navigating the complexities of IP rights in the age of AI requires a nuanced understanding of both technology and law, it's clear that collaboration between legal experts, technologists, and policymakers will be key to shaping a future where creativity and innovation flourish.
