AI laws and regulations

Overview of AI laws and regulations

Specific laws and regulations do not yet exist for the application of AI systems. However, some aspects are already governed by existing regulations:

Fundamental rights and freedoms

Rules for the protection of fundamental rights can be found in the Charter of Fundamental Rights of the European Union. In addition, there are many laws that deal specifically with one or more fundamental rights. For example, the General Data Protection Regulation (GDPR) prescribes specific rules guaranteeing the right to privacy and . The Dutch General Equal Treatment Act (Algemene Wet Gelijke Behandeling) establishes the prohibition of discrimination.

Liability

If one wants to use AI in a product, then the phenomenon of product liability is important. As explained in this article on liability for AI models, product liability can be used if something goes wrong with a physical product. Normal liability, such as a wrongful act (tort), is also explained in the aforementioned article. In addition to general forms of liability, there are of course specific rules, such as the Dutch Medical Devices Act (Wet Medische Hulpmiddelen) for medical devices in which AI is used.

Consumer protection

Consumers are currently better protected when buying physical products and services than when buying online "products" and services, including software. As we wrote earlier, the Dutch legislature must soon implement the new Digital Services and Products Directive. If this directive is in place, then consumers will also be well protected against non-conformity in online services as an AI assistant, such as failure to keep promises about what the AI assistant can do. In addition, the Netherlands also has yet to implement the Sale of Products Directive, which gives consumers a right for a certain period of time to the updates needed to keep a product fit for purpose.

Intellectual property

The source code of a computer program is eligible for copyright protection once there are sufficient creative choices in creating the software. To date, AI models have been treated as software, so this also applies to AI models. You can find more details in our article on copyright on AI models.

Security

The implementation of the European cybersecurity regulation is currently being debated. Once it comes into force, this regulation will introduce a harmonized system of certification that AI systems will probably also have to comply with.

Special AI legislation

The European Commission presented a proposal for the new AI Act on April 21, 2021. This regulation aims to regulate high-risk applications of AI. This proposal still has to go through the legislative process, and it will be a few years before there will be an AI Act. Unfortunately, we cannot clarify this yet. You can find a loose overview of the legislative process of the regulation here.

Guidance documents

The Personal Data Authority (AP) has published a document outlining its approach to overseeing AI and algorithms. Since the AP enforces under the GDPR, you can also find an overview of the relevant articles of the GDPR for the supervision of AI in the document's appendix.

In addition to the AP, the Dutch Consumer and Market Authority (ACM) also has a role in supervising AI and algorithms. The ACM does not focus on the processing of personal data. To clarify its position towards AI, the ACM has published a guidance document on possible influence of consumers by AI and also a Position Paper on how it plans to supervise algorithms that influence the market. 

The EU's High Level Expert Group on AI has published an Ethical Guideline on AI that explains, based on 7 principles, how to ethically design and use AI. In addition to this European guideline, there are also many local agencies and governments that have drafted their own ethical guidelines.