Mélanie Dulong de Rosnay est chargée de recherche au CNRS et directrice du Centre Internet et Société et du Groupement de Recherche Internet IA et société qui structure thématiquement la communaut. Elle travaille sur les biens communs numériques, la régulation par la technique, le droit et les politiques publiques des technologies de l’information.
Dr Andres Guadamuz is a Senior Lecturer in Intellectual Property Law at the University of Sussex and the Editor in Chief of the Journal of World Intellectual Property. His main research areas are on artificial intelligence and copyright, open licensing, cryptocurrencies, and smart contracts. Andres has published two books, the most recent one of which is "Networks, Complexity and Internet Regulation", and he regularly blogs at Technollama.co.uk. He has acted as an international consultant for the World Intellectual Property Organization, and has done activist work with Creative Commons.
Legal scholars have been studying whether AI- and computer-generated works have copyright protection. While the law about this type of creative work is answered satisfactorily in some countries such as the UK, the treatment of such works is less clear in other jurisdictions, and there is still no consensus as to whether AI works even have copyright in the first place.
Another legal debate addresses the use of preexisting human-generated, copyrightable works to feed and train machine learning models, towards the creation of those AI works: similarly to text and data mining, can this activity qualify as exception or fair use, or should original authors’ permission be obtained?
After a recap of the rules which may be applicable to both input and output creative works, the authors focus on the intersection between copyright and copyleft, the legal hack at the source of free, libre and open source software (FLOSS) and Creative Commons Share Alike licenses under which authors grant more freedoms to potential users than applicable copyright law, under the condition that their works and their derivatives will remain free for all to build upon.
Can machine learning programmers use Creative Commons licensed works to train their creation algorithm?
One of the open licensing controversy of the past was suggesting to introduce more restrictive moral clauses for free software, to limit the use of their work for military purposes, or by nazis. The debate resurfaced for surveillance industry as some proposed US Immigration and Customs Enforcement (“ICE”) or Palantir. And facial recognition programmers have been using photos available in open databases such as Flickr.
For non-software creative works, open licensing will authorize programmers to reuse Creative Commons -licensed works to train their AI.
What can we learn from copyleft?
In the same way Creative Commons included a restriction for non-commercial use, could there be an ethical or fair AI clause?