The European Commission concluded a call for evidence on its AI Strategy for the Cultural and Creative Sectors. Last week, COMMUNIA submitted its response (full text available as a PDF), highlighting two priorities: enabling the development of public-interest AI and safeguarding the public information ecosystem on which AI systems depend.
The strategy complements the Commission’s recently concluded call for evidence on a better copyright environment for European creativity and innovation (see our response). While some of the barriers we identify require legislative reform, the strategy can establish a broader vision for promoting access to culture, knowledge and information in the age of AI.
Ensuring access to high-quality data and legal certainty for public AI development
The cultural and creative sectors can support AI development by providing access to high-quality data. Yet several barriers prevent research and cultural heritage institutions from realising this potential:
- National cultural heritage laws can restrict the reuse of works and artefacts that are already in the Public Domain.
- Exclusive digitisation agreements keep around two million digitised Public Domain books from being made broadly available.
- Paywalls restrict access to publicly funded research.
- Legal uncertainty complicates reliance on the text-and-data-mining exception for scientific research and the sharing of datasets for verification and reproducibility.
The strategy should:
- make access to and reuse of data for public-interest purposes a clear EU priority;
- encourage Member States to remove barriers to reusing Public Domain heritage;
- promote open access to and open licensing of publicly funded research;
- provide legal guidance for researchers and public-interest AI developers;
- support standardised, machine-readable opt-out protocols; and
- prioritise a shared European training corpus comprising high-quality cultural and research materials.
Safeguarding the sustainability of the public information ecosystem
Commons-based knowledge projects and cultural heritage institutions provide the reliable information on which AI systems depend. These infrastructures are sustained largely through public funding, charitable support and voluntary contributions. At the same time, their operators face growing costs for hosting, preservation, digitisation, cybersecurity and managing AI-related traffic, while receiving little direct support from the commercial deployment of AI.
To keep these resources available, the strategy should:
- recognise the infrastructure costs of large-scale automated access;
- promote responsible crawling practices, including compliance with robots.txt and other relevant protocols;
- encourage AI providers to attribute and link to their sources; and
- explore an infrastructure fund that channels a share of the economic value generated by commercial AI to cultural heritage institutions and commons-based projects such as Wikipedia and OpenStreetMap.
