The European Writers’ Council represents 220,000 authors from 49 organisations and 31 countries. The EWC would like to raise three aspects:

(1) It’s only a machine – so do not over-humanise technical processes neither objectify human work!

Large language models are by no means intelligent, but technologies that generate mashups of protected works and individual expression from human authors. Books are not “data”, they are results of human creation. The terms “training” or “learning” humanise complex but entirely mathematical processes. Hence, we encourage to clearly distinguish terms as works created by human authors or reproduction processes.

(2) There is no Advanced Informatics without Artistic Intelligence – it’s all based on human author’s work!

All large language models only exist because they have digested the intellectual property, the works of human authors. Without human creation, there would be not a single generative software. Over 4 million books, most of them copyright protected, are exploited since at least a decade without authorisation, remuneration or transparent self-disclosure. There is an obvious need to prevent this unauthorised use of intellectual property.

(3) Artistic and authors’ intelligence need the ART principle:

Authorisation, Remuneration, Transparency.

Worldwide, writers are the sources of the entire book economy and create works at their own economic risk. Only the use of a work triggers a small financial share. The unauthorised and unremunerated communication of the mass use of human works by high-tech companies is causing huge damage to  creators’ markets. Opt-out is not an option; it shifts the burden of action and costs to authors. As Text and Data Mining is not the basis for any machine processing for generative AI, copying stays an exclusive right. Rights clearance and remuneration must be provided by companies.

Conclusion?

We consider world copyright law to be suitable for these challenges – it’s the enforcement which lacks ambition. Voluntary licensing, labelling and no copyright on machine output are the right way. Neither further limitations and exceptions nor a binding treaty or soft law should therefore be introduced.

“Training the Machines – Bytes, Rights and the Copyright Conundrum” presented by Nicole Pfister Fetz, Secretary General, onsite in Geneva (CH) in an exceptional speech at the WIPO Conference. EWC‘s Secretary-General pointed this aspects out and stood up for our interests.

… about the WIPO Conversations on Intellectual Property and Frontier Technologies

The WIPO Conversation on IP and Frontier Technologies provides an open, inclusive forum to engage with and facilitate discussion and knowledge building among the widest possible set of stakeholders on the impact on IP of frontier technologies, including artificial intelligence (AI).

WIPO generally holds two sessions of the Conversation each year in a format that allows the widest possible global audience to participate. Topics of the Conversation include the uses and applications of frontier technologies to assist IP Offices and IP owners as well as more conceptual policy based discussions.

Past sessions of the WIPO Conversation focused on GenAI, the Metaverse, data and frontier technologies in IP administration.

Full program of the 9th Session

Recording

Presentations: