AI and market abuse: do the laws of robotics apply to financial trading?

F. Consulich, M. Maugeri, C. Milia, T.N. Poli, G. Trovatore

Quaderno giuridico (Legal Research Paper) No. 29 - May 2023 [PDF]
 

Abstract
The article focuses on the distinction between weak AI systems and strong AI systems. While the former depends on preset instructions from manufacturers, programmers, or users, the latter has self-learning abilities and produces autonomous and unpredictable outputs compared to the initial inputs. The spread of such technologies in the financial market raises concerns about the adequacy of existing regula-tions, particularly about the liability of financial misconduct involving autonomous AI agents. While legal rules can be applied extensively to combat such misconduct for weak AI systems, ex novo criteria for responsibility attribution are needed for strong AI systems to make effective measures that protect the regular functioning of trading. The emergence of autonomous AI poses new protection needs in the face of a regulatory framework focused solely on human conduct. The study identifies three possible solutions aimed at repressing the conduct of AI systems that, autonomously and unpredictably, assume harmful or specifically market integrity-infringing behaviours. However, each of these solutions presents specific critical issues depending on the legal sectors involved as a result of non-human agents' illicit behaviour.  
[Text generated by a well-known online chatbot system, based on an advanced artificial intelligence system, in response to the authors' request (formulated at 13:23 on May 12, 2023) to produce an abstract of the "Conclusions" paragraph of this paper.]
 

Authors:
Federico Consulich - Università degli Studi di Torino
Marco Maugeri - Università Europea di Roma
Carlo Milia - CONSOB, Ufficio Abusi di Mercato (c.milia@consob.it)
Tommaso Nicola Poli - CONSOB, Ufficio Studi Giuridici (t.poli@consob.it)
Gianfranco Trovatore - CONSOB, Ufficio Studi Giuridici (g.trovatore@consob.it)

We would like to thank Francesca Medda and Filippo Annunziata for their comments. The opinions expressed in this Working Paper are solely attributable to the authors and do not represent the official positions of Consob, nor do they in any way engage the responsibility of the Consob. Therefore, when citing the contents of this paper, it is not correct to attribute them to Consob or its leadership. Any errors or inaccuracies are solely attributable to the authors. 

ISSN 2281-5236 [online]