EWC’s Ten Principles to regulate Generative AI

EWC Handout: ten tasks and future scenarios to be addressed by political decisionmakers and stakeholders to regulate generative AI.

The European Writers’ Council (EWC) is the world’s largest representation of writers in the book sector and of all genres (fiction, non-fiction, academic, children’s book, poetry, etc.).

With 49 organisations from 31 countries of the EU, EEA, and non-EU areas, the EWC represents 220.000 writers and translators.

As writers, illustrators, translators, narrators, as well as publishers, book trade and Collective Management Organisations are severely affected by the consequences of so-called (generative) AI, this Handout draft ten tasks and future scenarios to be addressed by political decisionmakers and stakeholders.

Every authorship needs protection under the ART principle: Authorisation – Remuneration – Transparency. There shall be no AI without ART.

1 Regulate Input: Unlicensed copying and other copyright infringements occur constantly at the INPUT level. GAI is largely based on years of unauthorised, unremunerated, and non-transparent use of protected works by often non-EU developers. Sanctions are needed.

2 Label Output: GAI products must be labelled without exception to avoid illegitimate compensation claims to CMOs, publishers or platforms, to allocate liability, to avoid unlawfully benefiting from public funding such as prizes, scholarships, or project funds.

3 Build a modern legal framework. Authors’ Rights and Copyright legislations must be, inter alia but not limited to, extended to new rights due to the GAI technologies, e.g., scraping; storage and reproduction for machine and GAI training; personal, moral, and authors’ right on one’s own style, voice, and appearance incl. sanctions against proximity within the GAI output.

Economic benefits from AI are an assumption, the creative economy is a fact that should not be carelessly sacrificed.

4 Stop disruptive transfers of economic value. Authors are the sources of a 22.5-billion-euro book economy in Europe. Their intellectual labour and thus the basis of a functioning market is being dismantled without remuneration to build a profitable competitive market through GAI applications, sold back in those countries whose intellectual resources have been plundered.

5 Protect the sensitive national ecosystem. To disrupt the thriving book market and its positive impact on employment, regional development, taxes, innovation driven economy and payments for social benefits, leads to long-term costly consequences for each State.

6 Protect literary diversity. Instead of licensing, the TDM exception for commercial purposes only offers an opt-out procedure as “protection” against GAI scrapers. The book sector is forced to implement highly costly administrative and technical mechanisms to protect their labour, which leads to lower investment capacity into authors and books. The framework for commercial TDM needs to be subject to licensing and remuneration, as opt-out requirement is unjustified for the true impact of scraping for machine learning.

7 Protect human authors from being ripped-off by GAI junk. Hundreds of thousands GAI outputs have been reaching the digital markets unlabelled for over a year. Fake authors, fake books with hallucinations or plagiarism, fake readers who push GAI output into bestseller lists, GAI translations without licence etc. redirect turnover revenues to illegitimate sources.

Defend democratic values and freedom of expression – also for the next generation

8 Don’t fall for the paradigm shift from man ./. machine: The uncritical enthusiasm for GAI and machine-made output, which is in fact the copying and remixing of human achievement, leads to a devaluation of human value in art and culture. This requires education about the production processes of GAI in opposite to human artistry, e.g., media education in schools.

9 Be aware of degeneration of individuality and diversity: The increasing use by children and young adults of GAI to produce culture-like outputs leads to the loss of cultural techniques in writing, visual arts, composing etc. This establishes an inter-generational inequity of learning to create art as free expression. Plus: GAI products focus on dominant languages, which dries up multilingualism, and reproduce and multiply white, masculine, Western perspectives. This leads to bias and intersectional discrimination.

10 Democracy at stake: the loss of trust in information and media due to falsified and unlabelled communication incl. voice cloning and deep fakes of persons or situations, is based on the growing inability to identify reality. The future of knowledge is at risk, also in connection with the educational mission of the state, and which can result in the undermining of democratic principles through growing malformation of public opinion.