Navigating AI together: the risks of plagiarism and protecting IP

By Sally Pritchett
CEO

In a thought-provoking session on AI, plagiarism and IP, we unravelled the challenges facing communicators using generative AI.

In our most recent ‘Navigating AI together’ session, we went on a journey to explore the legal challenges around the use of generative AI tools. We were delighted to be joined by Alex Collinson from Herrington Carmichael, a seasoned expert in commercial and intellectual property law.

The purpose of the ‘Navigating AI together’ series is to foster a community dedicated to responsible AI usage and to work through these challenges with clarity and integrity, together. In this recap of our discussion, we’re shining a light on the issues around AI and charting a path towards ethical and informed AI usage.

What do we mean by intellectual property, copyright and trademarks?

Many of us are likely guilty of using language like intellectual property, copyright and trademarks almost interchangeably. But to understand the legal landscape around AI, we need to get these terms nailed down.

Intellectual property (IP): refers to creations of the mind, such as inventions, literary and artistic works, designs, symbols, names, and images used in commerce. IP is protected by laws which give the creators or owners exclusive rights to use their creations or discoveries for a certain period of time.
Copyright: Copyright grants exclusive rights to creators of original works, such as books, music, and software, allowing them to control reproduction, distribution, and other uses for a limited time. Copyright acts as a deterrent to those who may wish to profit from reproduction, and can be licensed which can create a revenue stream for your business.
Trademark: Trademarks protect words, symbols, or designs that distinguish goods or services, providing exclusive rights to their owners to prevent confusion among consumers and maintain brand identity.

Common copyright misconceptions

Even before we get into the new, complex and unclear realm of generative AI, Alex shared with us that there are a lot of common misconceptions:

“Everything on the Internet is free to use”: Contrary to popular belief, just because something is freely available online doesn’t mean it’s in the public domain. Users must be cautious about where they obtain information and how they use it.
“My name is protected by copyright”: Names and symbols can’t be protected by copyright; instead, they fall under trademark law. Trademarks prevent others from using similar names or symbols that could cause confusion in the marketplace.
“You have to register your copyright to protect it”: Under English law, there’s no requirement to register copyright. Copyright protection automatically applies to works once they’re created, including artistic and literary works.

What the law says about generative AI

While the law is not quite up to speed with generative AI, Alex shared some interesting case examples to help us better understand the legal landscape.

Copyright

Alex shared with us a high-profile case, Getty Images (US) Inc v Stability AI Ltd [2023] EWHC 3090 (Ch), that highlights the difficulty generative AI is causing within copyright law. Getty Images alleged that Stability AI unlawfully scraped images from other websites to train its AI model, resulting in infringement of copyright and trademarks.

Despite Stability AI seeking summary judgment (without a full trial), the court refused, recognising the novel issues surrounding AI and copyright law. These legal nuances reveal the need for the law to adapt to the evolving landscape of AI technology, with more cases expected to emerge.

It is possible that copyright could be infringed when AI systems are trained, if this process involves copying a substantial part of copyright works that are still within the terms of protection. However, the issue regarding whether copying will be deemed to have occurred where a generative AI system has been trained on copyright works is not straightforward.

Potential issues include:

Proving the use of copyrighted works in training AI: proving that copyrighted works were used to train and develop a generative AI system poses challenges due to the lack of transparency in data usage.
Jurisdictional issues: determining the location of training and development activities, especially in cases involving online communication, can be complex. Although even if the activities occurred outside the UK but targeted the UK public, copyright law may still apply.
Permitted acts under copyright law: Assessing whether any infringement occurred involves considering permitted acts outlined in the Copyrights Act. These acts include fair dealing defences, such as criticism, review, quotation, parody, caricature, and pastiche, which provide exceptions to copyright restrictions.

Trademarks

Given how new generative AI is and the evolving legal landscape, there are very limited case studies available. However, Alex explained how longer existing AI applications may help us to understand potential future legal implications.

AI-driven recommendation systems, commonly used in e-commerce, analyse user data to suggest products based on preferences and behaviour. Amazon, for instance, employs such AI technology.

In the case of Cosmetic Warriors and Lush v Amazon ([2014]EWHC 181 (Ch)), Amazon was found to have used Lush products and branding on its website to attract customers and recommend other products without Lush’s consent. Lush claimed trademark infringement, asserting that its intellectual property rights were violated, with the court ruling in its favour. This case highlights the importance of protecting trademarks within the digital marketplace.

The ethical dilemma

While the law hasn’t yet adapted to the emergence of generative AI, as communicators, we can’t afford to sit back and wait while these tools that can enhance creativity and productivity are available to us. With the ongoing legal uncertainties, it’s up to us to evaluate the ethical implications of our communications practices.

Consider a scenario shared during our session, involving Keith Haring’s artwork and an AI-generated copy produced on Midjourney, a licensed platform. Keith Haring is perhaps most famous for his art used in the iconic Change4Life public health campaign.

The Keith Haring Foundation, a charitable organisation dedicated to children in need and those affected by HIV/AIDS, holds the rights to Haring’s iconic artwork.

However, we posed a hypothetical situation where the government wants to commission a new public health campaign, once again using Haring’s renowned illustrative style. As our Midjourney example shows, instead of licensing art from the Keith Haring Foundation, they could potentially use AI-generated images.

This capability raises ethical and legal concerns regarding intellectual property rights. Despite the AI’s capability to mimic Haring’s style, using these images without consent from the Keith Haring Foundation could potentially infringe upon their IP rights. Not to mention the ethical impact of not licencing the images from a charitable foundation.

This example reflects the importance of ethics when incorporating AI-generated content into communication strategies, while we wait for the law to catch up.

We’d like to thank Alex for guiding us through this complex topic. If you’d like to explore the legal implications of generative AI further, we encourage you to connect with Alex on LinkedIn to stay updated on future discussions and insights.