Let Kopiosto enforce your rights in AI development on your behalf

A new category of rights on the training and development of AI has been added to the Grant of Rights to Kopiosto. The new category of rights enables Kopiosto to monitor the rights of authors and other right holders when the works are used to train artificial intelligence and language models. The category of rights determines which rights Kopiosto can monitor for you and the types of use of works to which Kopiosto can grant licences on your behalf.

AI models and language models are trained on huge amounts of data, which include a very large amount of copyrighted material. So far, there has been no comprehensive effort to obtain licence from copyright holders to use this material for training AI or language models. As a creative professional, you must have the right to decide how your work can be used to train AI, and you must be properly compensated for the use.

By approving the addition to your Grant of Rights to Kopiosto, you enable Kopiosto to enforce your rights in the development of AI and to intervene in the unauthorised use of your works in the training of AI models, for example.

Approving the addition does not directly lead to the granting of licences for the use of your works in the training of AI.

After approval:

1. Kopiosto can enforce your rights and intervene in any unauthorised use for the training of AI models.
2. Kopiosto’s member organisations decide whether and under what conditions to develop licences for training AI software. The use of some material types may also be excluded from the licences.
3. Kopiosto may collect compensations on behalf of authors and distribute them to authors and other copyright holders in a manner approved by its member organisations.

Licences to use the works could be granted to large technology companies such as Google, OpenAI, Microsoft or Silo AI. Licences could also be granted to smaller start-ups or other actors developing AI or language models.

In practice, individual authors and publishers have no way of controlling or agreeing with large technology companies on the use of their works in the context of training AI and language models or commercial text and data mining. If there is a desire to create and enforce rules for the use of works in training AI models, and to compensate authors and other copyright holders for it, the only option is the collective management of rights. As a collective management organisation, Kopiosto will look after your interests in the negotiations.

By accepting the addition to your Grant of Rights to Kopiosto, you authorise Kopiosto to monitor the use of your works in the training of AI and language models as well as commercial text and data mining. This also allows Kopiosto to intervene in any unauthorised use of your works. By accepting the addition, you also enable Kopiosto to grant licences to use your works for commercial text and data mining as well as the training or development of AI and language models.

This allows you to retain control over the utilisation of your own works and receive compensation for their use in the training and development of AI and language models.

Please visit the Kopiosto authorisation service right away to approve the new category added to your Grant of Rights.

1. Log in to the authorisation service with your online banking credentials at https://kopiostoextra.fi/valtuudet.
2. There is an exclamation mark next to your Grant of Rights awaiting approval. Open the Grant of Rights by clicking View.
3. Click on the Update authorisation icon at the bottom of the Grant of Rights page.
4. Approve the updates to the categories of rights and press Save.

You will see an exclamation mark next to all of your pending Grants of Rights. Approve each of them individually.

Use a Chrome, Edge or Safari browser to approve the update to your Grant of Rights. Also make sure that you are using the latest browser version.

If you need more information or do not have access to the authorisation service, please contact Kopiosto, valtuudet(at)kopiosto.fi.

AI is trained with massive amounts of data. The data often also includes copyrighted content, such as texts and images collected from the web. For example, works can be scraped, i.e. collected from web pages, to form different datasets. These datasets, or masses of works, are used to train AI models and programs. Copyrighted works are also used for text and data mining, which is often included as a preliminary step in the process of training AI models.

The finished AI applications, in turn, use all the supplied material to further train and develop the application. This happens, for example, when a user of an AI application enters a copyrighted image into the application and asks the application to edit a new version of it. Users can also enter various texts, such as articles, research reports or excerpts from books, into the applications. The AI application is asked to provide e.g. summaries, translations or presentations of texts.

An AI application often refers to generative AI applications that produce content at the command or prompt of a human. The content produced by the application can be text, images or videos, for example. But AI is also used by other applications and software, such as more traditional word processors to proofread user-generated text, or by image processors, web search engines and recommendation algorithms.

An AI or language model is a machine learning model that aims to emulate human intelligence and the language, sound or images produced by humans, for example. The language model uses probabilities to predict things, such as the occurrence of words and the continuation of sentences, and to form question-answer pairs. The most sophisticated language models can also generate entire texts, such as articles, discussions or summaries of the text fed into the model. Similarly to texts, advanced AI models can also generate content, such as images or videos, based on user prompts.

AI models can be trained on any material available on the internet, using web scraping, text mining and data mining, among other procedures.

The material can also be obtained from a party managing the material, such as a library or archive collection or company database.

Moreover, AI applications can also access works from the material that users feed into them. Users can feed the application books, articles, pictures or music, for example.

So far, it is virtually impossible to know whether your work has been used to train AI. The majority of developers do not report the training material they use. Starting from the autumn of 2025, the EU AI Act will require developers of general-purpose AI systems and related service providers to produce a summary, which must indicate the training material used.

The non-profit organisation Common Crawl crawls the web and makes its archives and datasets freely available to the public. You can check the websites crawled by Common Crawl using Common Crawl Index Search. The organisation issues a report on the crawled websites on a monthly basis.

You can reserve your rights, i.e. prohibit the use of your works for commercial text and data mining in a machine-readable way, in writing through the terms of use of a website or service or a licence agreement, for example. You can also set a ban in the file metadata or use robots.txt or a similar restriction standard to prevent text and data mining relating to your work. You can ask for more information on this topic from Kopiosto’s lawyers, neuvonta(at)kopiosto.fi.

The update of the category of rights does not affect your agreement with your employer or your customers.

The Grant of Rights to Kopiosto covers the secondary use of copyrighted texts and images published in printed products or on the internet. Examples of such secondary uses include photocopying, scanning, and copying and printing online material for use in educational institutions as well as the internal use of public administration operators and businesses.

Thus, the authorisation you grant to Kopiosto with the Grant of Rights does not apply to the sale or publication of your works.

New categories of rights were also added to the Grant of Rights to Kopiosto in 2023. At the time, two completely new categories of rights were added and one category was extended.

The categories cover the use of works included in newspaper or magazine publications (GR8) and the sharing of works in online content sharing services, such as YouTube and Instagram (GR9). The extended category of rights, in turn, relates to the use of works included in the collections of libraries, archives and museums (GR5).

If you have not yet approved the update from two years ago, we recommend that you approve both updates at the same time.

Read more about the 2023 category of rights update.

On the front page of the authorisation service, there is an exclamation mark next to the Grant of Rights pending approval of the update. Once you have approved the update to your Grant of Rights, the exclamation mark will disappear and the message ‘Authorisation updated’ will appear in your Grant of Rights.

If the update cannot be approved, this is usually because you are using an unsupported web browser or your browser version is out of date. Use Chrome, Edge or Safari to update your Grant of Rights to Kopiosto. Make sure you have the latest browser version.

Your previous Grant of Rights to Kopiosto does not allow Kopiosto to grant licences for the use of works in the training or development of AI. Kopiosto needs new authorisations to enforce your rights and interests in the age of artificial intelligence.

You can easily approve an update to your Grant of Rights to Kopiosto through the Kopiosto authorisation service.

An exclamation mark will appear in the authorisation service next to any Grant of Rights pending your approval.

Kopiosto’s member organisations will be adopting the Grant of Rights containing the new category of rights at different times. You will receive a message from the organisation to which you have given a Grant of Rights to Kopiosto when the relevant update can be approved.

In other words, please note that you will need to approve each of your pending Grants of Rights separately.

You can check in the Kopiosto authorisation service whether you have provided a Grant of Rights to Kopiosto. If you have not yet issued a Grant of Rights to Kopiosto, you can do so through the authorisation service. The new Grant of Rights automatically includes the rights to train AI models.

Neither Kopiosto nor its member organisations can approve the update to your Grant of Rights on your behalf. The authorisation is a personal agreement that no one else can make or change on your behalf.

You can easily approve an update to your Grant of Rights to Kopiosto through the Kopiosto authorisation service. If you do not have access to the authorisation service, please contact Kopiosto, valtuudet(at)kopiosto.fi.