Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

När beviset inte kan litas på – En utredande uppsats om deepfakes och rättssäkerhet i svensk processrätt

Deuschl, Ebba LU (2026) JURM02 20261
Faculty of Law
Department of Law
Abstract
Technological development, particularly within AI, has enabled production of so-called deepfakes, i.e., artificially generated or manipulated material that may appear highly authentic. At the same time, digitalisation has made digital evidence central in legal proceedings. This creates new legal challenges, since convincing false material may undermine the reliability of digital evidence.
The purpose of this essay is to contribute to knowledge about how Swedish procedural law deals with evidence such as AI-generated and/or AI-manipulated deepfakes, and to analyse whether this framework is compatible with both formal and material legal certainty. Using a legal dogmatic method, the essay analyses deepfake evidence primarily in the light of... (More)
Technological development, particularly within AI, has enabled production of so-called deepfakes, i.e., artificially generated or manipulated material that may appear highly authentic. At the same time, digitalisation has made digital evidence central in legal proceedings. This creates new legal challenges, since convincing false material may undermine the reliability of digital evidence.
The purpose of this essay is to contribute to knowledge about how Swedish procedural law deals with evidence such as AI-generated and/or AI-manipulated deepfakes, and to analyse whether this framework is compatible with both formal and material legal certainty. Using a legal dogmatic method, the essay analyses deepfake evidence primarily in the light of the free consideration of evidence in Chapter 35, Section 1, first paragraph of the Swedish Procedural Code (SWE: 35 kap. 1 § första stycket RB). Peczenik’s definition of formal and material legal certainty is used as the theoretical starting point for the analysis of legal certainty.
The analysis shows that there is generally no obstacle to presenting deepfake evidence, under the principle of free production of evidence. Issues relating to deepfakes instead arise through objections by the parties or within the evaluation of evidence. Furthermore, analogies drawn from related practice show that there is no uniform method for handling suspected deepfake evidence. The legal position is therefore unclear. Deepfakes have also changed the conditions for the free evaluation of evidence, since it can no longer be assumed that audio, image and video evidence faithfully depict a real event. As a result, the assessment of digital evidence’s authenticity has become more difficult.
At present, there is no uniform method for the legal handling of deepfake evidence. The system therefore suffers from a lack of predictability, which is problematic from a formal legal certainty perspective. The essay argues that, under the current legal framework, deepfake evidence can be addressed either as a matter of burden of proof and evidentiary requirement, or as a matter of evaluation of evidence. Regardless of classification, however, significant issues of material legal certainty remain. If deepfake evidence is treated as a matter of burden of proof and evidentiary requirement, there is a risk that the assessment becomes far too unnuanced, while asymmetries between the parties may be reinforced. If it is instead treated within the evaluation of evidence, there is a significant risk of incorrect evidentiary evaluations. The current system therefore appears inadequate if the aim is to ensure legal certainty in disputes where deepfake evidence may occur. However, the essay does not argue for a legislative change of the free consideration of evidence. Instead, it concludes that deepfake evidence should be dealt with within the framework of the free evaluation of evidence, combined with further guidance on how such evaluation should be carried out, as well as measures to strengthen competence and improve access to appropriate tools. (Less)
Abstract (Swedish)
Teknikutvecklingen, särskilt inom AI, har möjliggjort produktion av deepfakes, det vill säga artificiellt genererat eller manipulerat material som kan framstå som mycket autentiskt. Samtidigt har digitaliseringen medfört att digital bevisning fått en allt mer central roll i rättsprocesser. Detta aktualiserar nya rättsliga utmaningar, eftersom möjligheten att skapa övertygande falskt material riskerar att urholka tillförlitligheten hos digital bevisning.
Syftet med uppsatsen är att bidra med kunskap om hur svensk processrätt hanterar bevisning i form av AI-genererade och/eller AI-manipulerade deepfakes samt att analysera om denna ordning är rättssäker i formellt och materiellt hänseende. Med en rättsdogmatisk metod analyseras... (More)
Teknikutvecklingen, särskilt inom AI, har möjliggjort produktion av deepfakes, det vill säga artificiellt genererat eller manipulerat material som kan framstå som mycket autentiskt. Samtidigt har digitaliseringen medfört att digital bevisning fått en allt mer central roll i rättsprocesser. Detta aktualiserar nya rättsliga utmaningar, eftersom möjligheten att skapa övertygande falskt material riskerar att urholka tillförlitligheten hos digital bevisning.
Syftet med uppsatsen är att bidra med kunskap om hur svensk processrätt hanterar bevisning i form av AI-genererade och/eller AI-manipulerade deepfakes samt att analysera om denna ordning är rättssäker i formellt och materiellt hänseende. Med en rättsdogmatisk metod analyseras deepfakebevisning särskilt i ljuset av principen om den fria bevisprövningen i 35 kap. 1 § första stycket RB, inklusive fri bevisföring och fri bevisvärdering. Som teoretisk utgångspunkt för rättssäkerhetsanalysen används Peczeniks definition av formell och materiell rättssäkerhet.
Analysen visar att det enligt den fria bevisföringen i regel inte föreligger något hinder mot att åberopa deepfakebevisning. Frågor om deepfakes aktualiseras i stället genom partsinvändningar och inom ramen för bevisvärderingen. Genom analogier från närliggande praxis framgår vidare att det saknas en enhetlig metod för att hantera misstänkt deepfakebevisning och att rättsläget därför är oklart. Deepfakes har också förändrat förutsättningarna för den fria bevisvärderingen, eftersom det inte längre på samma sätt går att utgå från att ljud-, bild- och videobevisning återger en verklig händelse. Bedömningen av digital bevisnings äkthet har därmed försvårats.
I dag saknas en enhetlig metod för juridisk hantering av deepfakebevisning. Ordningen präglas därför av bristande förutsebarhet, vilket är problematiskt ur ett formellt rättssäkerhetsperspektiv. I uppsatsen argumenteras för att deepfakebevisning enligt gällande rätt kan hanteras antingen som en bevisbörde- och beviskravsfråga eller som en bevisvärderingsfråga. Oavsett klassificering kvarstår dock betydande materiella rättssäkerhetsproblem. Om deepfakebevisning hanteras som en bevisbörde- och beviskravsfråga är risken att bedömningen blir alltför onyanserad, samtidigt som partsasymmetrier kan förstärkas. Om den i stället hanteras som en bevisvärderingsfråga finns en påtaglig risk för felaktiga bevisvärderingar. Den nuvarande ordningen framstår därmed som otillräcklig om ambitionen är att garantera en rättssäker prövning av tvister där deepfakebevisning kan förekomma.
Uppsatsen argumenterar inte för en lagändring av den fria bevisprövningen. I stället görs bedömningen att deepfakebevisning bör hanteras inom ramen för den fria bevisvärderingen, kombinerat med ytterligare vägledning om hur bevisvärderingen bör utföras samt kompetenshöjande åtgärder och ökad tillgång till ändamålsenliga verktyg. (Less)
Please use this url to cite or link to this publication:
author
Deuschl, Ebba LU
supervisor
organization
alternative title
When Evidence Cannot Be Trusted – An Investigative Essay on Deepfakes and Legal Certainty in Swedish Procedural Law
course
JURM02 20261
year
type
H3 - Professional qualifications (4 Years - )
subject
keywords
Civilrätt, Civilprocessrätt, Deepfakes, Deepfakebevisning, Fri bevisprövning, Rättssäkerhet, Formell Rättssäkerhet, Materiell rättssäkerhet.
language
Swedish
id
9226074
date added to LUP
2026-06-03 09:26:27
date last changed
2026-06-03 09:26:27
@misc{9226074,
  abstract     = {{Technological development, particularly within AI, has enabled production of so-called deepfakes, i.e., artificially generated or manipulated material that may appear highly authentic. At the same time, digitalisation has made digital evidence central in legal proceedings. This creates new legal challenges, since convincing false material may undermine the reliability of digital evidence. 
The purpose of this essay is to contribute to knowledge about how Swedish procedural law deals with evidence such as AI-generated and/or AI-manipulated deepfakes, and to analyse whether this framework is compatible with both formal and material legal certainty. Using a legal dogmatic method, the essay analyses deepfake evidence primarily in the light of the free consideration of evidence in Chapter 35, Section 1, first paragraph of the Swedish Procedural Code (SWE: 35 kap. 1 § första stycket RB). Peczenik’s definition of formal and material legal certainty is used as the theoretical starting point for the analysis of legal certainty.
The analysis shows that there is generally no obstacle to presenting deepfake evidence, under the principle of free production of evidence. Issues relating to deepfakes instead arise through objections by the parties or within the evaluation of evidence. Furthermore, analogies drawn from related practice show that there is no uniform method for handling suspected deepfake evidence. The legal position is therefore unclear. Deepfakes have also changed the conditions for the free evaluation of evidence, since it can no longer be assumed that audio, image and video evidence faithfully depict a real event. As a result, the assessment of digital evidence’s authenticity has become more difficult.
At present, there is no uniform method for the legal handling of deepfake evidence. The system therefore suffers from a lack of predictability, which is problematic from a formal legal certainty perspective. The essay argues that, under the current legal framework, deepfake evidence can be addressed either as a matter of burden of proof and evidentiary requirement, or as a matter of evaluation of evidence. Regardless of classification, however, significant issues of material legal certainty remain. If deepfake evidence is treated as a matter of burden of proof and evidentiary requirement, there is a risk that the assessment becomes far too unnuanced, while asymmetries between the parties may be reinforced. If it is instead treated within the evaluation of evidence, there is a significant risk of incorrect evidentiary evaluations. The current system therefore appears inadequate if the aim is to ensure legal certainty in disputes where deepfake evidence may occur. However, the essay does not argue for a legislative change of the free consideration of evidence. Instead, it concludes that deepfake evidence should be dealt with within the framework of the free evaluation of evidence, combined with further guidance on how such evaluation should be carried out, as well as measures to strengthen competence and improve access to appropriate tools.}},
  author       = {{Deuschl, Ebba}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{När beviset inte kan litas på – En utredande uppsats om deepfakes och rättssäkerhet i svensk processrätt}},
  year         = {{2026}},
}