Digital resilience podcast series | Episode 1 | Intellectual property and data privacy issues in artificial intelligence

Following on from our successful Digital Transformation series, this is the first episode of our new Digital Resilience podcast that focuses on managing risks in a digital world. This first instalment looks at the intellectual property (IP) and data privacy issues that arise from the use and deployment of artificial intelligence (AI) solutions. Our series host Dino Wilkinson is joined by Chris Williams, Partner in Clyde & Co’s intellectual property disputes practice in London, and Masha Ooijevaar, a Senior Associate in Dubai who advises clients on data protection and cybersecurity.

The episode begins by exploring issues of personal privacy in the context of AI solutions and how far current data protection laws apply to the application and use of personal data in AI. The discussion moves on to consider the IP issues relevant to  AI deployment, including authorship of AI works and who is responsible for potential breaches. Finally, participants discuss the challenges for regulators and how organisations can best manage and mitigate risk in such a fast-moving landscape.

With AI now capable of collecting, processing, and analysing huge volumes of information, Ooijevaar starts by outlining the data privacy issues for individuals in connection with both the input material used to train AI systems and the generated outputs of AI solutions. Developers of AI systems must consider this when building the technology and organisations using AI tools must also be aware of this issues. In addition, Ooijevaar says that regulators around the world are closely scrutinising the situation, balancing “how to protect individual rights while also supporting opportunities for innovation and economic growth.”  

On the topic of current regulations, Ooijevaar explains that while it may not always be clear which laws apply, some existing guidance and concepts will certainly be relevant. For example, the concept of ‘privacy by design’ and the practice of conducting privacy impact assessments are equally valuable in an AI context.

Williams outlines the main IP issues concerning AI, which similarly impact both the inputs and outputs of the technology. While it is clear that the information ingested by AI systems may be subject to IP protection, it is perhaps less clear who, if anybody, owns the work created by an AI system.

On the first issue, Williams advises companies to take steps to minimise IP disputes: “Do not assume that any or all material in your possession - or, indeed, publicly available - is free to upload,” he says. Equally, companies should take steps to protect their own material being used by AI, by limiting their availability or putting contractual safeguards in place.

When it comes to AI authorship, Williams describes this as “certainly uncertain”, and dependent on several factors, including originality and the level of automation involved. While UK copyright laws recognise that “the person who made the necessary arrangements for the creation of the computer-generated work is taken to be the author,” he says this may not necessarily apply in the case of AI, which “increasingly applies its own skill and judgment.”

A rapidly changing environment means that governments are monitoring the situation closely with regulators such as the UK Information Commissioner’s Office releasing guidance on what businesses need to consider. Williams says that governments must “look to weigh up the interests of the creative community on the one hand and the technology companies on the other.” He notes that case law will also help to clarify the situation, providing additional guidance.

In the meantime, as change is happening so quickly, he advises businesses to take a pragmatic approach “rather than doing nothing and waiting for this mystical framework to emerge by legislation or case law, because that may not happen, certainly for the immediate term.”

Listen to the rest of the episodes in the series