Legislation in the age of innovation: regulating AI-driven child sexual abuse material in the European Union. Fact or fiction?
Legislation in the age of innovation: regulating AI-driven child sexual abuse material in the European Union. Fact or fiction?
Date
2024
Authors
Bergh, Cézanne : Van den
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis, titled ‘Legislation in the Age of Innovation: Regulating AIDriven
Child Sexual Abuse Material (AI CSAM) in the European Union (EU)
– Fact or Fiction?’, sheds light on the complexity of regulating AI CSAM in
the EU – a minefield of legal, ethical, and practical deficiencies. It reveals the
imperative to address these intertwined conundrums, which are essential to
achieve effective EU regulation of AI CSAM. This material, comprising digitally
manipulated content of real children and AI-generated fictitious CSAM,
perpetuates real CSAM through Generative AI (GenAI) models’ training data
and weights. This blurring line between real and AI CSAM compels the EU
to deepen its understanding and develop more effective legal strategies.
The current EU legislative landscape, including the CSA Directive, the
proposed CSAM Regulation, and the AI Act, overlooks the intricacies of AI
CSAM, rendering it ill-equipped to combat its creation and dissemination.
Additionally, regulating AI CSAM aligns poorly with general EU principles
and key criminal law requisites, such as criminal intent, identifiable victims,
and causation of real harm. Extending the criminal focus to GenAI models
and their owners further complicates the fit within the traditional framework.
These legislative impediments pose grave ethical hazards, normalising child
sexual abuse, obstructing criminal investigations, (re-)victimising children,
and escalating financial sextortion. Given the severe infringement upon child
dignity, integrity, privacy, wellbeing, and protection, as well as their best
interests, effective practical legal solutions are urgently needed. Yet, current
regulatory obligations for GenAI models and online platforms are practically
limited, given their circumvention and ineffectiveness in detecting AI CSAM.
Therefore, a more intrusive, paradigm-shifting approach, including expanded
criminal accountability, could enhance practical effectiveness. However, its
authoritarian implications, conflicting with EU human rights and democratic
values, undermine its practical feasibility. This spurs further research to
explore innovative ways to combat these legal, ethical, and practical hurdles
impeding effective EU regulation, while maintaining a pragmatic outlook.
Description
Second semester University: Lund University. Awarded thesis 2023/2024.
Keywords
child abuse,
sexual abuse,
artificial intelligence,
European Union,
law