Artificial intelligence and copyright

Q: How do technology companies justify the unauthorised use of works in the development of artificial intelligence?

Technology companies, particularly large platforms and artificial intelligence operators, 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 these exceptions apply comprehensively to the entire process, from training artificial intelligence to content generation. The explanatory notes to the section of the Directive concerning text and data mining do not take a position on whether that section applies to the operation of artificial intelligence. The matter will be settled in court proceedings, several of which are also underway in Europe.

Q: Why are artificial intelligence and copyright sparking debate right now?

Generative AI is developing and becoming more widespread rapidly, but the rules governing it are still being worked out. Copyrighted works have also been used extensively to train AI models without any agreement having been reached with the copyright holders regarding such use, or any transparent disclosure having been made regarding the training materials. The use of works requires transparent practices, authorisation from copyright holders and appropriate remuneration. Solutions already exist, such as in the form of collective licensing, but their implementation requires cooperation between AI companies, the creative industries and legislators.

Q: How is artificial intelligence utilised in the creative industry?

The creative industry uses artificial intelligence for activities such as brainstorming, translating texts and producing summaries and abstracts. In a survey carried out by Kopiosto in 2025, 57% of the creators who responded said that they had used artificial intelligence in their creative work. More than half of the respondents predicted that its use would increase in the future, particularly in the automation of manual tasks, proofreading and the development of ideas.

Generative AI applications train their models by using creative works, but there are no common practices in place and no agreement has been reached with the authors or other copyright holders regarding the use of these works. This poses a challenge to the realisation of copyright and the operating conditions of the creative economy in the age of artificial intelligence.

How is copyright realised in the training of artificial intelligence?

At present no agreement has been made with the authors regarding the use of their works for training artificial intelligence. Furthermore, developers of artificial intelligence do not comply with authors’ reservation of rights, meaning prohibitions on the use of their works.

Furthermore, there is a lack of transparency regarding the materials used for the training, which makes it practically impossible for authors to monitor the enforcement of their rights or grant licences for the use of their works.

Generative AI models have been, and continue to be, trained by using copyrighted content, such as literature, visual arts and music. Copyright means that the author or other copyright holder of a work has the exclusive right to decide on – i.e. to authorise or prohibit – the use of their work, and to receive appropriate remuneration for this.

How can copyright and the development of artificial intelligence be reconciled?

Reconciling artificial intelligence and copyright generally requires developers of artificial intelligence to seek the permission of copyright holders to use copyrighted works in the training and operation of AI and to pay appropriate remuneration for such use. In this way, professionals in the creative industry are involved in the AI value chain and receive appropriate remuneration for the use of their works.

Copyright organisations can offer collective licences that enable the use of works in the development of artificial intelligence in a comprehensive manner, with the copyright holders’ consent and in return for remuneration. This would also provide legal certainty for companies developing artificial intelligence. However, technology companies avoid obtaining licences, and resolving the situation requires legislative support.

Why is transparency in artificial intelligence important?

To ensure compliance with copyright and the reservation of rights, the training materials used for artificial intelligence models must be transparent. When developers openly disclose the materials used to train the models, copyright holders can grant the necessary licences for the use of the content. Transparency builds trust between the creative industry and AI companies.

In Kopiosto’s view, responsible artificial intelligence is built on innovation, respect for copyright, the transparency of training materials and an appreciation of creative work. The aim is to develop artificial intelligence that benefits users, creators and society as a whole.

Finland cannot build its AI future at the expense of creative work. If works are used to train artificial intelligence without permission or compensation, this will undermine the economic foundations of the cultural and content sectors.

CEO of Kopiosto Valtteri Niiranen