Ready for AI - The legal issues you need to be aware of
Artificial intelligence, AI, is on everyone's lips and has the potential to transform entire industries and society at large. At the same time, the rapid pace of development and the far-reaching risks involved mean that it is essential that this work is carried out in a safe, legal and responsible manner.
‘‘All AI solutions should be developed and used within the framework of the law, and stakeholders should be held accountable accordingly. We strongly advocate reasonable and relevant transparency regarding how AI systems are developed and function,‘’ writes AI Sweden, which coordinates Swedish initiatives in this area.
But how can companies navigate this landscape without making mistakes or jeopardising future development opportunities? We asked two experts at CMS Wistrand, Jennie Nilsson, partner specialising in AI, GDPR and information security, and Erik Ullberg, partner specialising in AI, tech, market and IP law. They will also be participating in the AI Industrial Summit organised by Ny Teknik on 27 May to highlight the legal perspective on AI.
What kind of AI-related services are your clients asking for these days?
‘‘Typical compliance projects that many companies need help with involve formulating procedures and policies for using and assessing the risks of AI systems in their businesses. For clients who are suppliers of AI systems, our services may include helping with documentation of the system and how to respond to questions from their customers. In connection with acquisitions, it is important to understand how far the company being acquired has come with its compliance.’‘
‘‘We also help with training, as there is a knowledge requirement in the new AI Act. This simply means that people who work with AI must have sufficient knowledge to do so. We also provide advice on risks specific to Sweden, as risk assessments may differ between countries depending on guidelines and decisions made by different authorities. CMS is currently represented worldwide with over 80 offices.‘‘
How early in the AI development process should legal advice be sought to minimise risk?
‘‘The only good answer to that question is as early as possible, so that requirements under the AI Act, GDPR and intellectual property laws are considered from the outset. It is not only Swedish laws and regulations that you need to comply with. The use of data must not violate any laws or agreements. Data itself is not protected unless it is covered by copyright. You must ensure that you allow for its use in your agreements and that you do not inadvertently waive the right to use data for purposes such as training an AI tool, and that you secure any rights that may be necessary.‘‘
What challenges arise when AI and GDPR meet?
‘‘The definition of personal data is so broad under the GDPR that it poses challenges for how personal data can be used to train AI models. If you use personal data, there are quickly a lot of requirements that must be met, and you should seek help if the expertise is not available within the company. We also recommend working in parallel with the GDPR and the AI Regulation when launching new AI systems, as this can lead to significant efficiency gains. In addition, you should make sure to stay up to date on the practices of GDPR supervisory authorities, as they have been active in several countries specifically regarding GDPR and AI.‘‘
And what challenges do you see in the intersection of AI and IP rights?
‘‘The most obvious risk is if copyright-protected material is used in AI. This will inevitably involve disposal within the meaning of copyright law, which requires the consent of the copyright holder. Another challenge is if you give generative AI instructions and process existing protected data, with the risk of plagiarism or infringement. This naturally raises the question of who is responsible for an infringement: is it the provider of the AI tool, the person who writes the prompt, or the person who uses it?''
‘‘There are a number of court cases ongoing across Europe, but we are probably far from seeing the end of the tunnel here. As the use of AI increases, so too will the number of conflicts and copyright disputes.‘‘
There are a number of court cases ongoing across Europe, but we are probably far from seeing the end of the tunnel here. As the use of AI increases, so too will the number of conflicts and copyright disputes
Erik Ullberg
How do you, as a lawyer, keep up with the rapid development of AI?
‘‘We make sure to stay up to date on the measures being taken by various regulatory authorities, attend events, engage in CMS's international network for those working with AI, and maintain extensive contacts with market players who deal with the AI Regulation in different ways. Developments are happening fast, but they don't always go straight ahead; they shift back and forth a bit, so it's also about being agile.‘‘
Source: Report by CMS: Technology Transformation. Survey on over 500 corporate counsel and risk managers from multiple industries around the world.
