Generative AI and the law: how do you stay compliant?

Bas Alderding

Generative AI is hot. Models such as GPT-4 and Stable Diffusion impress with their ability to generate text, images and other content. But what about the legislation surrounding the use of this technology? In this blog, we dive into the main legal concerns and provide tips on how to stay compliant.

The EU AI Act: first legislation for AI

The European Union is leading the way when it comes to regulating AI. In December 2023, the European Parliament and the Council agreed on the AI law. This law introduces rules and obligations for developing and applying AI, based on the level of risk:

  • Unacceptable risk: prohibited practices such as social scoring

  • High risk: strict requirements for recruitment, education and law enforcement, among others

  • Limited risk: transparency obligations such as labelling of AI-generated content

  • Minimal risk: free use allowed for e.g. video games and spam filters

For generative AI, 'limited risk' rules are especially relevant. Users should be informed if they are dealing with an AI system. And AI-generated text, images and videos should be labelled as such if they are intended to inform the public.

Copyright and data protection

In addition to the AI Act, there are two main legal areas of concern: intellectual property and data protection.

Generative AI models are trained on huge datasets of text, images, etc. By no means all this material is copyright-free. Using copyrighted material to train an AI model may infringe copyright. There is still a lot of uncertainty about this. Lawsuits such as Getty Images v Stable Diffusion will hopefully bring more clarity.

Privacy is also an important issue. Personal data should not simply be used to train AI models. The AVG has strict requirements for processing personal data. Anonymised or synthetic data sometimes offer a solution.

Tips compliant use of generative AI

So how do you make sure your use of generative AI is compliant? Some tips:

  • Be transparent. Inform users clearly when dealing with an AI system. Label AI-generated content.

  • Know the risks. Determine which risk category your application falls into according to the AI Act. Comply with the corresponding obligations.

  • Respect rights. Do not use copyrighted material or personal data to train your AI model unless you have permission.

  • Stay tuned. Technology and regulations are evolving. Make sure you know what's going on.

  • Ask for advice. In doubt about whether your application complies with the rules? Then seek legal advice. Better safe than sorry.

With the right knowledge and precautions, you can take full advantage of the opportunities of generative AI. Just make sure you respect the law. This way, you will contribute to the responsible application of this fascinating technology.

Do you have questions about the legal aspects of AI? Then get in touch with us. We are happy to think along with you!

SevenLab is a reliable party with great customer focus. They actively think with you, are enormously helpful and flexible.

Satish Bahwanidin

Project leader VO-raad

Ready to discuss your project with us?

Meet the SevenLab Team and Founders