LAW FIRM FOR EU AI ACT COMPLIANCE

Definition of AI

Artificial Intelligence (also called AI) is to be achieved by algorithms and software techniques and especially by machine learning. The classical machine learning consists of the methods based on classical statistics and probability theory. The newer so-called neural networks include the so-called deep learning and reinforcement learning. While classical software is rule-based, machine learning software systems are data-based. The prerequisite for functioning Artificial Intelligence is therefore sufficiently large data volumes of sufficient quality. Application areas of Artificial Intelligence are, for example, autonomous driving or robotics.

Copyright infringements by AI

Copyrights can be infringed in the context of AI at different levels. For instance, the unauthorised use of copyright protected material for training AI systems can constitute a copyright infrigement. Another example can be the use of protected material in the outcome of the AI system.

Personal data violation by AI

Use of personal data by AI systems without suffient consent or justified purpose runs the risk to be considered a breach of data privacy laws – both use as training data and use as output of AI. Also the collection of personal data provided by the user can violate the law.

EU AI Act

The EU regulation on Artificial Intelligence (EU AI Act) is intended to further promote the development and use of Artificial Intelligence. It aims to ensure a safe and trustworthy use of Artificial Intelligence by containing potential risks of Artificial Intelligence.

The EU AI Act has come into force on 2 August 2024 and will continue to become applicable stepwise. After 2 February 2025 the next deadline will be on 2 August 2025, the last is scheduled for 2 August 2027.

The text of the EU AI Act can be found here:

https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689

The EU AI Act applies to providers worldwide placing on the EU market or putting into service AI systems in the EU or placing on the EU market general-purpose AI models. And it applies to professional deployers of AI systems that have their place of establishment or are located within the EU and to professional deployers of AI systems the output of which is used in the EU.

The AI Act classifies AI systems in different categories, in accordance with their respective risk. The main compliance obligations with regard to high-risk systems are:

– establishment of risk management system;

– data and data governance

– technical documentation;

– record keeping;

– transparency and information;

– human oversight;

– accuracy, robustness and cybersecurity.

Also from AI Systems with limited risks will follow a number of compliance obligations, like:

– quality management

– documentation keeping

– automatically generated logs

– corrective actions and duty of information

– Cooperation with competent authorieties

Specific obligations are imposed on the developers and deployers of general purpose AI models.

Law firm assists developers and providers of AI systems to comply with EU AI Act

In a first step, the EU AI Act addresses providers of AI Systems. Depending on the risk of the respective AI System, the providers have to comply with specific obligations. Thus, above all, the right category needs to be confirmed. This is not only a technical but also a legal task. Following from the categorization, the individual obligations need to be determined. Getting from the abstract determination of the obligations to the specific action in practice is likely to require further legal assistance.

Law firm support AI providers in out-licensing

Providers of AI models or systems should monetize their develpments on the basis of a solid licensing agreement. We help you to negotiate an appropriate deal and enter into a valid and enforceable agreement.

Users of AI systems can easily qualify as deployer or even provider

From the use of an AI system as such compliance obligations do not necessarily follow. However, a user, for instance, can qualify as provider if it applies its trademark to the system or substantially modifies the system.

Law firm assists companies to in-license AI Systems

Integrating AI in your company’s workflow requires also legal preparation. We help you to enter into licensing contracts with providers of AI systems – from the first negotiations to execution of the license agreement. Where adaptions of the system to your individual requirements are necessary we support this process by precisely phrasing the obligations for the developers and by backing the implementation of the instructions.

Law firm assists companies with compliant use of AI Systems

For compliant use of an AI system – or a general purpose AI model – it needs a categorisation of both the respective system and the role of the respective use. Even if the use is considered compliant at the beginning, by its actions the user can change the categories. For example, altering the purpose of deploying an AI system can change the categorisation – so that “limited risk” can turn into “high risk”. Or by modifying technical specifications, the simple use may becom a provider in the sense of the EU AI Act.