Copyright lessons on Machine Learning:
what impact on algorithmic art?
by Theodoros Chiou*
© 2019 Theodoros Chiou
Everybody may disseminate this ar ticle by electronic m eans and make it available for downloa d under the terms and
conditions of the Digital P eer Publishing Licence (DPPL). A copy of the license text may be obta ined at http://nbn-resolving.
Recommended citation: Th eodoros Chiou, Copyright lessons on Machine Lea rning: what impact on algorithmic art? 10 (2019)
JIPITEC 398 para 1
Keywords: artificial intelligence; machine learning; text and data mining; algorithmic art; copyright; copyrighted
works; Infosoc Directive; DSM Directive; Big Data; reproduction right; adaptation right; copyright
Machine Learning workflow typically involves the re-
alization of (multiple) reproductions of any protected
work used as training material. The present paper
aims to assess the extent to which the use of copy-
righted works for Machine Learning purposes in the
field of algorithmic creativity is controlled by the mo-
nopolistic power of the copyright rightholder on that
work. The answer to this question will be researched
in the context of EU copyright law, by examining the
content of reproduction right and exceptions possibly
applicable in a typical ML workflow in the field of al-
gorithmic art, before making an overall assessment
of the current EU regulatory framework for artistic
ML projects, as it is shaped after the DSM Directive
Abstract: Nowadays, Artificial Intelligence (AI) is de-
scribed as “the new electricity”. Current algorithmic
innovation allowed the development of software
which enables machines to learn and to achieve au-
tonomous decision making, with limited or no human
involvement, in a vast number of applications, such
as speech recognition, machine translation and algo-
rithmic creation of works (computer generated art),
on the basis of a process widely known as Machine
Learning (ML). Within the ML context, machines are
repeatedly trained by means of specifically designed
learning algorithms that use a corpus of examples in
the form of data sets as training material. Very often
and, especially in the context of algorithmic creativ-
ity, the training material is mainly composed by copy-
righted works, such as texts, images, paintings, musi-
cal compositions, and others.
The objective of Making machines intelligent.
Articial Intelligence may be seen from different
standpoints and receive accordingly different
interpretations. From a rather technical point of
view1, Articial intelligence is the eld of computer
* Dr. Theodoros Chiou is Post-Doc Researcher at the University
of Athens, School of Law (Department of Private Law)
and Attorney-at-law (IPrights.GR). Email: Theodoros.
email@example.com. This paper is based on a conference
presentation delivered by the author during the 9th ICIL
Conference, “Psychological and socio-political dynamics
within the Web: new and old challenges to Information Law
and Ethics”, held in Rome, Italy, July 11-13, 2019.
1 For a different approach, see among others Stuart Russell
& Peter Norvig, Articial Intelligence: A Modern Approach (3rd
ed., Pearson 2010) 1: “the study of agents that exist in an
environment and perceive and act”.