Ett nytt verktyg, samma plikter - En analys av generativ AI i advokatverksamhet i ljuset av god advokatsed
(2025) LAGF03 20252Department of Law
Faculty of Law
- Abstract (Swedish)
- Enligt 8 kap. 4 § rättegångsbalken ska en advokat redbart och nitiskt utföra anförtrodda uppdrag och iaktta god advokatsed. Bestämmelsen speglar advokatyrkets särskilda ansvar och aktualiserar bland annat krav på tystnadsplikt och diskretion, omsorg och kompetens samt lojalitet. Uppsatsen behandlar hur advokatetiska krav påverkas när generativ artificiell intelligens (GAI) används som arbetsverktyg i advokatverksamhet.
Uppsatsens syfte är först att undersöka vilka advokatetiska problem som kan uppstå vid användning av GAI, särskilt i relation till tystnadsplikt och diskretion, omsorgsplikt och kompetens, lojalitetsplikt och intressekonflikter samt krav på oberoende. Därefter analyseras i vilken utsträckning det nuvarande advokatetiska... (More) - Enligt 8 kap. 4 § rättegångsbalken ska en advokat redbart och nitiskt utföra anförtrodda uppdrag och iaktta god advokatsed. Bestämmelsen speglar advokatyrkets särskilda ansvar och aktualiserar bland annat krav på tystnadsplikt och diskretion, omsorg och kompetens samt lojalitet. Uppsatsen behandlar hur advokatetiska krav påverkas när generativ artificiell intelligens (GAI) används som arbetsverktyg i advokatverksamhet.
Uppsatsens syfte är först att undersöka vilka advokatetiska problem som kan uppstå vid användning av GAI, särskilt i relation till tystnadsplikt och diskretion, omsorgsplikt och kompetens, lojalitetsplikt och intressekonflikter samt krav på oberoende. Därefter analyseras i vilken utsträckning det nuvarande advokatetiska ramverket är tillräckligt för att hantera dessa problem och om det finns behov av förtydliganden eller vidare normutveckling.
Genomgången visar att GAI framför allt aktualiserar risker knutna till kontrollen över klientinformation och åtkomst, särskilt när uppgifter hanteras i system där datalagring, vidareanvändning eller tredjepartsåtkomst kan förekomma. Vidare framträder kvalitets- och tillförlitlighetsproblemen. Verktygen kan producera övertygande men felaktiga uppgifter och källhänvisningar, vilket skärper kravet på kontroll och verifiering innan material används i rådgivning eller processhandlingar. Uppsatsen belyser även lojalitets- och konfliktaspekter, där risken ligger i att klientintressen inte hålls tillräckligt separerade eller att bedömningar påverkas av bristande överblick över informationsflöden. Slutligen behandlas risken för överberoende och ”automation complacency”, vilket kan urholka advokatens självständiga analysarbete.
GAI inte är oförenlig med god advokatsed i sig, men advokatetisk förenlighet är villkorsstyrd. För att plikterna ska kunna upprätthållas krävs att användningen sker under kontrollerade former, att klientinformation hanteras under advokatens kontroll, att AI-utdata behandlas som preliminärt material som systematiskt granskas samt att det slutliga professionella omdömet och ansvaret för bedömningen förblir advokatens.
Sammantaget bedöms det nuvarande advokatetiska ramverket vara tillräckligt på principnivå, men det finns behov av förtydliganden på tillämpningsnivå för att öka förutsebarheten vid GAI-användning. (Less) - Abstract
- Pursuant to Chapter 8, Section 4 of the Swedish Code of Judicial Procedure, Swedish lawyers (members of the Swedish Bar) must perform entrusted assignments conscientiously and with integrity and comply with good professional conduct. This requirement reflects the special responsibilities of the legal profession and, inter alia, engages the duty of confidentiality and the duty of discretion, the duty of professional competence and the duty of loyalty. This thesis examines how these duties are affected when generative artificial intelligence (GenAI) is used as a work tool in legal practice.
The thesis has two aims. First, it identifies the ethical problems that may arise when GenAI is used in legal services, with particular focus on duty... (More) - Pursuant to Chapter 8, Section 4 of the Swedish Code of Judicial Procedure, Swedish lawyers (members of the Swedish Bar) must perform entrusted assignments conscientiously and with integrity and comply with good professional conduct. This requirement reflects the special responsibilities of the legal profession and, inter alia, engages the duty of confidentiality and the duty of discretion, the duty of professional competence and the duty of loyalty. This thesis examines how these duties are affected when generative artificial intelligence (GenAI) is used as a work tool in legal practice.
The thesis has two aims. First, it identifies the ethical problems that may arise when GenAI is used in legal services, with particular focus on duty of confidentiality and the duty of discretion, the duties of care and professional competence, the duty of loyalty and conflicts of interest, and the requirement of independence. Second, it assesses the extent to which the current ethical framework is sufficient to address these problems and whether clearer guidance or further normative development is required.
The analysis shows that GenAI primarily creates risks relating to the control of client information and access, especially where data may be stored, reused, or accessed by third parties within vendor chains. GenAI also raises quality and reliability concerns, as such systems may produce convincing but incorrect or fabricated content, increasing the need for verification before outputs are used in legal advice or procedural documents. In addition, the thesis high-lights loyalty and conflict-of-interest issues, particularly where the separation of client matters may be undermined by opaque information flows. Finally, the risk of over-reliance on GenAI may weaken independent professional judgment and accountability.
The thesis concludes that GenAI is not inherently incompatible with good professional conduct. However, ethical compliance is conditional. To uphold professional duties, GenAI must be used under controlled conditions, client information must remain under the lawyer’s control, AI-generated output must be treated as preliminary material subject to systematic review and the final legal assessment and responsibility must remain with the lawyer.
Overall, the existing ethical framework is largely sufficient at the level of principles, but further clarification at the level of application is needed to ensure predictability in GenAI use. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9217654
- author
- Eriksson, Amanda Madeleine Claire LU
- supervisor
- organization
- course
- LAGF03 20252
- year
- 2025
- type
- M2 - Bachelor Degree
- subject
- keywords
- AI, Generativ AI, Advokatetik, God advokatsed, Procedure, Processrätt
- language
- Swedish
- id
- 9217654
- date added to LUP
- 2026-02-09 14:58:59
- date last changed
- 2026-02-09 14:58:59
@misc{9217654,
abstract = {{Pursuant to Chapter 8, Section 4 of the Swedish Code of Judicial Procedure, Swedish lawyers (members of the Swedish Bar) must perform entrusted assignments conscientiously and with integrity and comply with good professional conduct. This requirement reflects the special responsibilities of the legal profession and, inter alia, engages the duty of confidentiality and the duty of discretion, the duty of professional competence and the duty of loyalty. This thesis examines how these duties are affected when generative artificial intelligence (GenAI) is used as a work tool in legal practice.
The thesis has two aims. First, it identifies the ethical problems that may arise when GenAI is used in legal services, with particular focus on duty of confidentiality and the duty of discretion, the duties of care and professional competence, the duty of loyalty and conflicts of interest, and the requirement of independence. Second, it assesses the extent to which the current ethical framework is sufficient to address these problems and whether clearer guidance or further normative development is required.
The analysis shows that GenAI primarily creates risks relating to the control of client information and access, especially where data may be stored, reused, or accessed by third parties within vendor chains. GenAI also raises quality and reliability concerns, as such systems may produce convincing but incorrect or fabricated content, increasing the need for verification before outputs are used in legal advice or procedural documents. In addition, the thesis high-lights loyalty and conflict-of-interest issues, particularly where the separation of client matters may be undermined by opaque information flows. Finally, the risk of over-reliance on GenAI may weaken independent professional judgment and accountability.
The thesis concludes that GenAI is not inherently incompatible with good professional conduct. However, ethical compliance is conditional. To uphold professional duties, GenAI must be used under controlled conditions, client information must remain under the lawyer’s control, AI-generated output must be treated as preliminary material subject to systematic review and the final legal assessment and responsibility must remain with the lawyer.
Overall, the existing ethical framework is largely sufficient at the level of principles, but further clarification at the level of application is needed to ensure predictability in GenAI use.}},
author = {{Eriksson, Amanda Madeleine Claire}},
language = {{swe}},
note = {{Student Paper}},
title = {{Ett nytt verktyg, samma plikter - En analys av generativ AI i advokatverksamhet i ljuset av god advokatsed}},
year = {{2025}},
}