The increasing use of generative artificial intelligence (gen AI) in law enforcement raises unprecedented challenges for criminal procedure and evidentiary theory. This paper investigates the deployment of deepfakes and other synthetic contents by police authorities or public investigative actors during the pre-trial phase, focusing on the implications of deceptive AI practices for the relationship between facts, evidence, and criminal adjudication. In particular, the paper examines how the use of synthetic images, voices, and audiovisual contents as tools of investigative interaction and deception may shape suspect behaviour and contribute to eliciting confessions, which may subsequently become central elements within the evidentiary record, even when the synthetic content itself does not enter the trial file. The analysis explores whether the use of gen AI for deceptive purposes in criminal investigations may be considered unlawful under the current EU legal framework. Particular attention is devoted to the AI Act, especially in relation to prohibited AI practices (Art. 5) and transparency obligations (Art. 50), as well as to the data protection framework applicable to law enforcement activities, i.e., the Directive (EU) 2016/680 and Art. 8 of the EU Charter of Fundamental Rights. ECHR and the case-law of the Strasbourg Court is considered as the general framework, due to Art. 6 TEU. The normative status of such practices lies at the core of the analysis, wondering whether they should be understood as a prohibited investigative method or, alternatively, as an emerging tool of criminal investigation. Against this backdrop, the paper aims to offer an exploratory analysis of the legal and theoretical implications of (‘investigative deepfake’) and to open a broader discussion on their admissibility, legitimacy, and consequences within the EU.

Gen AI for Deceptive Purposes in Pre-trial Investigations. Deepfakes vis-à-vis Human and Fundamental Rights

Paseri, L.
;
Quattrocolo, S.
2026-01-01

Abstract

The increasing use of generative artificial intelligence (gen AI) in law enforcement raises unprecedented challenges for criminal procedure and evidentiary theory. This paper investigates the deployment of deepfakes and other synthetic contents by police authorities or public investigative actors during the pre-trial phase, focusing on the implications of deceptive AI practices for the relationship between facts, evidence, and criminal adjudication. In particular, the paper examines how the use of synthetic images, voices, and audiovisual contents as tools of investigative interaction and deception may shape suspect behaviour and contribute to eliciting confessions, which may subsequently become central elements within the evidentiary record, even when the synthetic content itself does not enter the trial file. The analysis explores whether the use of gen AI for deceptive purposes in criminal investigations may be considered unlawful under the current EU legal framework. Particular attention is devoted to the AI Act, especially in relation to prohibited AI practices (Art. 5) and transparency obligations (Art. 50), as well as to the data protection framework applicable to law enforcement activities, i.e., the Directive (EU) 2016/680 and Art. 8 of the EU Charter of Fundamental Rights. ECHR and the case-law of the Strasbourg Court is considered as the general framework, due to Art. 6 TEU. The normative status of such practices lies at the core of the analysis, wondering whether they should be understood as a prohibited investigative method or, alternatively, as an emerging tool of criminal investigation. Against this backdrop, the paper aims to offer an exploratory analysis of the legal and theoretical implications of (‘investigative deepfake’) and to open a broader discussion on their admissibility, legitimacy, and consequences within the EU.
2026
Progress in Artificial Intelligence. EPIA 2026.
Springer
Lecture Notes in Computer Science
17198
246
259
978-3-032-44475-2
Paseri, L.; Quattrocolo, S.
File in questo prodotto:
File Dimensione Formato  
2026_Paseri&Quattrocolo_EPIA.pdf

Accesso aperto

Tipo di file: PDF EDITORIALE
Dimensione 741.23 kB
Formato Adobe PDF
741.23 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2318/2169952
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact