At the end of 2023, the Federal Council decided to harness the potential of artificial intelligence (AI) while minimising the risks for society. To this end, it commissioned DETEC to draw up an overview of possible regulatory approaches to AI, with a deadline at the end of 2024. Since then, various interest groups have formed to make their voices heard in politics when it comes to possible AI regulation.
Federal Council wants to regulate AI – we want to have our say
SUISA is not alone in positioning itself either. As an association of Swiss collective management organisations, Swisscopyright stands up for copyright and neighbouring rights and for the interests of rightsholders. Swisscopyright has now drawn up a position paper in order to contribute to the public discourse.
In the position paper, the Swiss collective management organisations demand, among other things, that the copyright relevance of the use of protected works for the purpose of AI algorithm training be considered in any regulation, and that authors should therefore participate in the revenues of generative AI systems in the same way as is customary in the commercial licensing business. The full text of the position paper follows.
Swisscopyright: Position on the treatment of generative AI and its output from a copyright perspective
As an association of Swiss collective management organisations (ProLitteris, SSA, SUISA, Suissimage and Swissperform) Swisscopyright stands up for copyright and neighbouring rights and for the interests of rightsholders. In this position paper, we set out our position and demands on the regulation of artificial intelligence (AI). The current Swiss Copyright Act is to be used as a foundation to be built on. The collective management organisations can serve as central licensing bodies for certain uses by generative AI systems.
Literary and artistic works are intangible assets that are assigned to an owner by law. Copyright protection incentivises creativity, promotes the creative sector and guarantees cultural diversity. However, legal incentives become ineffective if comparable products created by generative AI compete with works created by humans. The result would be a dysfunctional, unbalanced creative market. This would ultimately pave the way for market failure.
That is something that must be prevented. AI regulation in the field of culture should therefore be based on the following principles:
- Copyright incentivises human creativity and ensures broad access to works and services. The Swiss Copyright Act (URG) is a suitable working basis, even under AI conditions. However, if a revision of the law were to be initiated, it would have to be ensured that a future law also follows the principle of human creativity being remunerated. Any exceptions to copyright and royalty-free limitation provisions should be reduced or consistently supplemented with the right to receive remuneration.
- A protection system is useless if it only works in theory. Changes are needed in practice when it comes to acquiring rights of use to protected works and performances by generative AI systems. There is a need for compliance obligations (clarification and observance of rights) and information on works and services used, and finally an obligation to label the use of generative AI systems.
- The decline in revenue suffered by the owners of copyrights and neighbouring rights because of artificially generated products must be counterbalanced and compensated. Revenues from the AI systems serve as the basis for the remuneration to be negotiated for the training of generative AI systems. Authors must share in the profits from generative AI systems in the same way as is customary in the commercial licensing business.
- Any regulation must reflect the copyright relevance of training with pre-existing works and services by AI providers. It must stipulate that co-operation with the collective management organisations is required. This applies not only to the legalisation of AI input, but also to the transparent transmission of usage data and the technically usable transfer of data. For reasons of practicability, the mass utilisation that takes place can only be managed collectively (by collective management organisations).
Since effective and efficient licence models are of central importance, the collective management organisations, as reliable and seasoned users of copyright law in practice, must be involved in the development of the regulatory environment. Collective management organisations ensure that the rights and interests of authors and holders of neighbouring rights are protected. We call on all stakeholders to define future framework conditions together with us practitioners. Then, innovation is promoted, and existing rights are safeguarded.
