2023 has been a year of generative AI developments. Easy access to AI tools has revolutionized industries and promised increased efficiency for all. With new opportunities, however, come new and challenging questions. How does copyright law apply to AI-generated works? Can AI be an inventor or an author? What are the contractual and ethical issues involved in using data to train AI models?  

Caravel’s Director of Legal Innovation, Monica Goyal and IP lawyer, Alex Stack explored the intersection of AI and Intellectual Property law in a live event. Sharing their various expertise, Monica and Alex explored points of conflict between AI and IP and the questions that these conflicts have prompted. Caravel will be hosting a second event for the public on February 9th, you can register for our Exploring AI and IP Law event here! 

Intellectual property (“IP”) covers various types of legal rights that protect the creations of the human mind. Some of the main types of IP are patents, which protect inventions; copyrights, which protect original works of authorship; and trademarks, which protect distinctive signs that identify the source of goods or services. IP is important for innovation, as it provides incentives and rewards for creators, and facilitates the dissemination and sharing of knowledge. 

In contrast, the appeal of AI is its ability to perform tasks that would otherwise require human intelligence, such as learning, reasoning, and decision-making. While AI is working to transform and revolutionize most fields, these efforts complicate IP law as it attempts to navigate issues of creation, innovation, and originality with AI.  

AI as an Inventor or Author: 

As new technologies emerge with the capability to create images, designs, and content, people using these tools have also sought to get copyright over those works. One of the main questions that arises in the context of AI and IP is whether AI can be an inventor or an author of a patentable invention or a copyrightable work.  

This question challenges the traditional notion of IP as a human creation and raises various legal, ethical, and policy implications. For example, can an AI system that generates novel ideas and inventions be listed as the sole inventor of a patent application? Or can an AI system that generates original works of authorship, such as GPT-3, be recognized as the author of a copyright registration? These questions have been tested in several countries, including Canada and the US, where patent and copyright offices have rejected the claims that AI can be an inventor or an author, on the grounds that the inventor or the author must be a natural person, not a machine.  

Patent and copyright offices have relied on the statutory language, the case law, and the policy rationale of the IP system, which all assume that the inventor or the author is a human being who conceived of the invention or the expression. However, some argue that denying IP rights to AI-generated inventions or works would discourage innovation and that the law should adapt to the changing realities and challenges of AI.  

Others argue that many AI tools still require “skill and judgement” from a human, to achieve the desired output. In which case, the AI is a tool utilized by a human to create, rather than being the creator themselves, and as long as the creation reflects the skill and judgment of the human, the human has a copyright in the creation.   

Ownership of AI-Generated Outputs: 

Another question that arises in the context of AI and IP is whether the use of AI to generate new outputs based on data or inputs infringes the IP rights of the owners of the data or inputs. For example, can the use of copyrighted works, such as books or images, to train an AI system, such as GPT-3 or DALL-E, be considered a fair use or a fair dealing, or does it require a license from the copyright owners? Can the generation of new works, such as code or images, that are similar to the works used to train the AI system, such as Copilot or Stability AI, be considered a transformative or a derivative work, and/or does it constitute a copyright infringement?  

Once again, these questions highlight the legal issues that arise from the intersectionality of IP and AI. Copyright owners have filed lawsuits, attempting to sue AI companies for using their works without permission or attribution, and for distributing the works generated by the AI system that they claim infringes on their copyright rights. There are legal issues with the arguments they make, as was discussed by Alex and Monica in their session. Primary questions include whether there is even copying of the works in question, whether the AI companies had legal access to the allegedly infringing works, and whether such copying (if proven) constitutes fair use of the work.  It is worth noting that use of a copyrighted work of itself is not an infringement.  

These questions raise complex and novel issues about the role and status of AI in the IP system and the rights and responsibilities of the human actors involved in the creation and use of AI. The IP issues in AI are further complicated by the myriad of proposed regulations such as the proposed Canadian Artificial Intelligence and Data Act, which may or may not weigh in on the issue of IP infringement with AI. These are not only legal concerns but also ethical and policy-oriented, as they touch upon the values and goals of the IP system, and the impact of AI on society and culture.  

There is no clear or uniform answer to these questions, different jurisdictions continue to take different approaches and perspectives. We can expect that as the capability of generative AI adapts, so will these policies. Caravel’s live event on February 9th with Monica Goyal and Alex Stack will continue to explore the issues of AI and IP in more depth. Register for the event here! 

Caravel has a team of over 100 qualified and experienced lawyers, including those specializing in IP. Get in touch with our team today to find out more.