Kopiosto’s solutions: How can Finland become a model country for ethical AI?

While artificial intelligence is in many ways a transnational issue, we must not simply give up and sit back waiting for solutions from the EU. Finland must act as a trailblazer and use legislation to strengthen the enforcement of copyright in the AI environment.

Finland can become a pioneer in ethical AI if the development of AI brings together innovation, respect for copyright, the transparency of training materials and an appreciation of human creative work. Responsible artificial intelligence serves users, creative professionals and society as a whole.

We need to establish rules for artificial intelligence that will build trust within society. Finnish people must be able to use generative AI that is not only efficient but also responsible. Kopiosto is ready to play its part in building responsible AI.

CEO of Kopiosto, Valtteri Niiranen

We have listed five solutions that could help make Finland a pioneer in ethical AI:

1. A licence and compensation as prerequisites

Generative AI has been, and continues to be, trained by using works by creative professionals without their permission or any compensation. In Kopiosto’s view, creative professionals must be part of the AI value chain. Permission must be obtained from authors to use their works for training artificial intelligence, and they are also entitled to remuneration.

Technology companies, particularly large platforms and artificial intelligence operators, continue to avoid licensing and paying compensation to copyright holders. These actors often invoke the exceptions relating to text and data mining set out in the EU’s DSM Directive. However, it is open to interpretation whether those exceptions apply comprehensively to the training of artificial intelligence and the use of works in generative AI. There are currently several court cases underway on this matter around the world.

2. Training materials must be transparent

From the perspective of responsible artificial intelligence, it is crucial to know what data was used to train the language model behind the AI. This is necessary in order for the copyright owners of the works to be able to enforce their rights.

Transparency regarding the training materials builds trust between the creative industry and AI companies. If no information is provided regarding the sources used to train artificial intelligence, Kopiosto takes the view that the so-called presumption of use should apply. This means that, unless the AI developer proves otherwise, it is presumed that copyrighted content has been used in the training and development of the AI.

3. A responsible national language model

Generative AI is based on a language model that has been trained on a large volume of works and data. Our benchmark countries, Norway and the Netherlands, have decided to develop their own national language models, which are trained with high-quality data in a responsible manner. This means that the language model is trained by using a wide variety of copyrighted content, for which permission has been obtained and for which remuneration is paid to the copyright holders.

In Kopiosto’s view, Finland should be a model country for ethical artificial intelligence. This would require AI to be based on a national language model that has been produced and subsequently trained in a responsible manner, with its training carried out in a way that respects copyright.

4. Cracking down on deepfakes

Artificial intelligence makes it possible to misuse individuals’ identifiable characteristics without their consent. Deepfake videos or images can be created of public figures, politicians or even private individuals, using their likeness, voice or bearing without their permission.

In Finland, there is no specific protection for a person’s likeness, voice or bearing. In Kopiosto’s view, Finnish legislation and legal remedies must be brought into line with technological developments so that identity theft and damage to people’s reputation can be effectively prevented.

5. The use of artificial intelligence must be mentioned

Artificial intelligence and other technologies can be tools for creating content, but they must not replace human creativity and personal input in the final result. Copyright and related rights arising from human creative work must always be recognised as belonging to a human being.

AI-generated content should therefore be clearly labelled. The EU’s Artificial Intelligence Act (2024/1689) also stipulates that the use of AI must be disclosed, at the very least, in situations where people might be misled into believing that the content is genuine or created by a human.