Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Between Regulatory Fixity and Flexibility in the EU AI Act

Larsson, Stefan LU ; Hildén, Jockum LU and Söderlund, Kasia LU (2024)
Abstract
The EU AI Act aims to regulate artificial intelligence (AI) in a way that balances innovation and protection from harms, but faces the challenge of keeping pace with the fast and dynamic development of AI. This paper examines the tension between fixity and flexibility when regulating emerging technologies, drawing on literature on the so-called pacing problem, contrasted by sociolegal theory on the importance of predictability and legal certainty, as a basis to address recent changes in the regulatory AI landscape. Specifically, it analyses how the EU AI Act employs various mechanisms of flexibility, such as i) voluntary measures and codes-of-conduct, ii) delegated and implementing acts, and iii) harmonised standards to cope with the... (More)
The EU AI Act aims to regulate artificial intelligence (AI) in a way that balances innovation and protection from harms, but faces the challenge of keeping pace with the fast and dynamic development of AI. This paper examines the tension between fixity and flexibility when regulating emerging technologies, drawing on literature on the so-called pacing problem, contrasted by sociolegal theory on the importance of predictability and legal certainty, as a basis to address recent changes in the regulatory AI landscape. Specifically, it analyses how the EU AI Act employs various mechanisms of flexibility, such as i) voluntary measures and codes-of-conduct, ii) delegated and implementing acts, and iii) harmonised standards to cope with the uncertainty and complexity of AI – potentially at the expense of predictability. The study therefore focuses primarily on how the AI Act addresses the emergence of general-purpose AI and generative AI, to illustrate challenges associated with regulating rapidly developing technologies. In conclusion, the paper argues that while flexibility is unavoidable when drafting law explicitly targeting such a swiftly moving and conceptually blurry field and concept as AI, it also entails trade-offs such as reduced legal predictability, which is concerning since predictability is essential for ensuring trust and legal certainty in the regulatory framework around this set of technologies. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Book/Report
publication status
submitted
subject
keywords
AI Act, The pacing problem, legal certainty, legal flexibility, general purpose AI, delegated acts, harmonised standards
pages
17 pages
project
The Automated Administration: Governance of ADM in the public sector
AI-förordningen - mellan fixering och flexibilitet
AI Transparency and Consumer Trust
language
English
LU publication?
yes
id
b05aadc4-b6f3-4789-a575-3d1e7627ea16
date added to LUP
2024-02-18 14:00:09
date last changed
2024-02-20 08:58:56
@techreport{b05aadc4-b6f3-4789-a575-3d1e7627ea16,
  abstract     = {{The EU AI Act aims to regulate artificial intelligence (AI) in a way that balances innovation and protection from harms, but faces the challenge of keeping pace with the fast and dynamic development of AI. This paper examines the tension between fixity and flexibility when regulating emerging technologies, drawing on literature on the so-called pacing problem, contrasted by sociolegal theory on the importance of predictability and legal certainty, as a basis to address recent changes in the regulatory AI landscape. Specifically, it analyses how the EU AI Act employs various mechanisms of flexibility, such as i) voluntary measures and codes-of-conduct, ii) delegated and implementing acts, and iii) harmonised standards to cope with the uncertainty and complexity of AI – potentially at the expense of predictability. The study therefore focuses primarily on how the AI Act addresses the emergence of general-purpose AI and generative AI, to illustrate challenges associated with regulating rapidly developing technologies. In conclusion, the paper argues that while flexibility is unavoidable when drafting law explicitly targeting such a swiftly moving and conceptually blurry field and concept as AI, it also entails trade-offs such as reduced legal predictability, which is concerning since predictability is essential for ensuring trust and legal certainty in the regulatory framework around this set of technologies.}},
  author       = {{Larsson, Stefan and Hildén, Jockum and Söderlund, Kasia}},
  keywords     = {{AI Act; The pacing problem; legal certainty; legal flexibility; general purpose AI; delegated acts; harmonised standards}},
  language     = {{eng}},
  month        = {{01}},
  title        = {{Between Regulatory Fixity and Flexibility in the EU AI Act}},
  url          = {{https://lup.lub.lu.se/search/files/171573074/Larsson_Hild_n_S_derlund_Jan_2024_Between_Regulatory_Fixity_and_Flexibility_in_the_EU_AI_Act_draft_paper_2024-1-26.pdf}},
  year         = {{2024}},
}