The intellectual property saga: approaches for balancing AI advancements and IP protection |Part 3

The first part on AI and IP discussed the complexities of copyrighting AI-generated content, noting the challenges of traditional laws in ownership. The second essay explored AI’s impact on trade secrets and trademarks in the EU and US legal frameworks. In this concluding section, methods being used to protect intellectual property in the era of AI will be explored.

Understanding AI and IP together is tricky. Unlike traditional forms of intellectual property, such as patents or copyrights, AI-generated outputs raise questions about ownership and authorship. Consequently, devising robust strategies to delineate ownership and protect AI-generated creations remains a major concern. As AI technology advances, it challenges traditional notions of ownership and attribution, asking for a re-evaluation of existing IP laws and ethical considerations. One significant aspect of this future is determining who owns the rights to AI-generated creations. For instance, if an AI system autonomously composes a symphony or designs a groundbreaking invention, should the credit and ownership belong to the programmer who developed AI, the company that deployed it, or perhaps even AI itself? This question suggests the need for clarity in IP laws to incentivise investment in AI research and was discussed thoroughly in the first essay. In patent cases, AI systems currently lack legal recognition as inventors, which could lead to legal reforms to accommodate AI-generated inventions. Similarly, copyright laws require adaptation to address ownership issues surrounding AI-generated creative works. Meanwhile, in trademark law, questions arise regarding the licensing and authorisation of AI systems for trademark use. As discussed in the second essay on this topic, many AI innovators choose trade secret protections over patents due to the ambiguity in traditional laws regarding AI and copyright. This approach allows them to keep their AI advancements confidential, making it challenging for others to detect and replicate their innovations, especially when used commercially. 

Legal protection for AI products?

Legal battles, such as Thaler vs Vidal, where Thaler filed patent applications for two inventions attributed to the DABUS AI without human involvement, illustrate the struggle to define AI’s role in intellectual property (IP) law. Typically, humans contribute to AI development, and its knowledge base includes copyrighted material. For instance, the US Court of Appeals ruled against recognising AI as an inventor, emphasising human-centric patent laws. Similarly, copyright registrations for AI-generated works face rejection due to human authorship requirements. 

However, cases like Thaler vs Perlmutter and Kashtanova’s comic book registration required protection for human-authored components of AI-generated content. The US Copyright Office faced this when Kristina Kashtanova sought registration for a comic book made with Midjourney AI. The US Copyright Office allowed Kashtanova to copyright the text and the arrangement of text alongside AI-generated artwork. While the text was deemed a product of human creativity, the registration also protected the arrangement. However, it explicitly excluded copyright for the AI-generated artwork itself. 

Patenting AI systems hurdles are further demonstrated under Alice Corp. vs CLS Bank International. This case established a two-step test for patent eligibility. The system assessed whether a patent claim involves ineligible subject matter like abstract ideas. If so, it considers whether the invention adds an ‘inventive concept’ to make it eligible. With this, many software-based and algorithm-reliant patents have been deemed ineligible. Given AI’s reliance on software and algorithms, inventors must navigate Alice carefully when patenting AI-related innovations or deciding whether to pursue patents at all, as it is not clear how much these rules will affect patents for software and AI. 

Approaches for IP Protection in AI

A study from 2023 from the University of Zurich’s Center for Intellectual Property and the Swiss Intellectual Property Institute proposes clarifications for AI-related IP aimed to provide clarity on this matter. The project suggests recognising AI as ‘inventors’ for patent protection, while human authorship takes precedence for copyright. Copyright may be granted for content jointly created by AI and humans, provided that human creativity is evident. Furthermore, companies should be able to claim ownership of AI-generated IP, without the need for new IP rights. In addition, permissive protection is advised to prevent AI owners from facing lawsuits for unintentional IP infringements.

Distinguishing between inspiration and infringement is crucial when asking for the establishment of governance mechanisms to address these concerns and maintain trust within creative industries. Recent conflicts, such as those between the Writers Guild of America (WGA) and the Alliance of Motion Picture and Television Producers (AMPTP), show the need for governance in creative AI usage. 

The negotiations between the two parties in this case included demands to restrict AI’s involvement in content creation, though compromises were reached to balance innovation with copyright protection. The agreement does not prohibit the use of AI but places restrictions on how it is credited and utilised. It states that neither traditional AI nor generative AI can be considered ‘writers’ or ‘professional writers’, and material produced solely by AI is not recognised as literary material. However, the agreement allows for collaborative work between writers and AI tools, with studios aiming for copyrightable material resulting from human-AI collaboration. Safeguards are also in place to ensure that AI use does not compromise copyrightability, with companies retaining the right to reject AI use if it affects copyrightability or work exploitation.

Detecting AI infringements

According to Originality.AI, an AI detection tool, almost 20% of the top 1000 websites in the world block crawler bots from collecting web data for AI use. Large language models (LLMs) such as OpenAI’s GPT family and Google’s LaMDA family require massive amounts of data to train their AI systems. Subsequently, various technology providers have developed and now offer AI-powered solutions designed to assist businesses in monitoring and protecting their intellectual property online. These solutions use machine learning algorithms to analyse vast amounts of data and detect potential instances of infringement. They provide tools for tracking the use of copyrighted material across websites, social media platforms, and digital channels, enabling rights holders to take appropriate action to protect their IP rights.

In August 2023, OpenAI launched its GPTBot crawler aiming to gather data for enhancing future AI models. Major websites (including Amazon, Quora, The New York Times, CNN, ABC, Reuters, and many others) have taken proactive measures to block AI crawlers from accessing their content. Axel Springer and the Associated Press have recently signed an agreement with OpenAI to license its news content for training AI models. Crawlers work like web browsers but save data instead of showing it. They are used by search engines like Google to collect information. While site owners can instruct crawlers to avoid their site, compliance is voluntary, leaving room for noncompliance by malicious actors. Google and other internet companies view the activities of their data crawlers as fair use. However, numerous publishers and holders of intellectual property have voiced objections to this, leading to several lawsuits against the company.

Government intervention through legislation is further aiming to enhance IP protection in AI. Legislative bodies play a critical role in introducing safeguards against the unchecked use of AI in accessing and utilising copyrighted material, thereby safeguarding the interests of creative industries in the digital age. For instance, a report from the UK Culture, Media and Sport Committee, composed of MPs from different parties, criticised the policies of the current UK administration, pointing out flaws and expressing concerns. In particular, they objected to the initial proposal to exclude text and data mining from copyright protection, suggesting it reveals a lack of comprehension of the creative industry’s importance to the economy and its employment of millions. 

In September 2023, lawmakers in France introduced a draft bill to regulate how artificial intelligence interacts with copyright laws. The aim was to make sure that AI respects creators’ rights, gets proper permission to use their work, and gives them fair credit. The proposed changes to the French Intellectual Property Code would mean that AI needs permission before using copyrighted material. Additionally, the law suggests a new tax on companies using AI to create works with uncertain origins. 

In the USA, lawmakers and regulatory bodies have also been struggling with the implications of AI on intellectual property rights. In response to the rapid advancements in generative AI and its widespread adoption, the US Copyright Office is reviewing the copyright implications. This action follows requests from Congress, the public, creators, and users of AI technology. Additionally, the US Patent and Trademark Office (USPTO) has examined the patentability of AI inventions and issued guidance for patent examiners. The guidance addresses the complexity of identifying substantial human input in AI-assisted inventions, providing key principles. According to it, merely posing a problem to an AI system is usually not significant, but crafting prompts tailored for specific solutions might demonstrate contribution. While acknowledging an AI-generated outcome as inventive does not automatically confer inventor status, making substantial contributions to it might. 

Looking ahead 

Effective regulation of intellectual property rights concerning AI systems and their creations is crucial not only for legal clarity but also for motivating innovation in the market. Given the novelty of AI-generated artistic works, a reevaluation of current approaches to this issue seems unavoidable. Key points could include regulations of IP rights for AI systems and their creations. A likely solution involves implementing a distinct protection system for AI-generated creations, with rights held by either the AI system’s creator or user based on specific criteria. Additionally, further discussions might be needed to address the protection of algorithms, which are currently not covered under the existing EU legislative framework.