JUNE 2023

EUROPEAN
LAWYERS IN
THE ERA OF
CHATGPT
Guidelines on how
lawyers should take
advantage of the
opportunities offered by
large language models
and generative AI

The European Bars Federation

(Fédération des Barreaux d’Europe)
New Technologies Commission
FBE REPORT // JUNE 2023

02

TABLE OF CONTENT

3 Introduction

4 Overview

5 Definitions

6 Guideline 1

7 Guideline 2

8 Guideline 3

9 Guideline 4

10 Guideline 5

11 Guideline 6

12 Guideline 7

13 Summary

14 Call to Action
FBE REPORT // JUNE 2023

03

INTRODUCTION
HOW LAWYERS SHOULD BENEFIT FROM LARGE
LANGUAGE MODELS AND GENERATIVE AI

As the New Technologies Commission of

the European Bars Federation, our goal is
to ensure that European lawyers are well-
versed in new technologies, enabling them
to use AI tools responsibly and in
alignment with professional principles.

The discussion regarding the impact of artificial

intelligence (AI) on the legal sphere has been
ongoing for some time. However, the introduction
of OpenAI's ChatGPT on November 30, 2022,
served as a significant catalyst for exploring how
generative AI tools could transform legal
services. As of June 2023, we are still in a phase
where general-purpose AI chatbots are
accessible, but it is expected that specialized
generative artificial intelligence systems will
emerge over time, specifically designed for the
legal field, aiming to streamline tasks for legal
professionals.
FBE REPORT // JUNE 2023

04

OVERVIEW
OUTLINE OF GUIDELINES ADDRESSING THE USE
OF GENERATIVE AI BY ATTORNEYS

The FBE's New Technologies Commission

recommends the following guidelines on the
applications of large language models and
generative artificial intelligence tools based on
them in legal practice:

1. Understand the generative AI technology

2. Acknowledge limitations and context
3. Adhere to existing rules on AI use
4. Complement legal expertise
5. Maintain attorney-client privilege
6. Ensure data protection and privacy
7. Inform clients and assume responsibility

While these guidelines may not be exhaustive,

they should be taken into consideration in order
to maintain ethical standards, protect client
confidentiality, and ensure responsible and
informed usage of generative artificial
intelligence and large language models in the
legal field.
FBE REPORT // JUNE 2023

05
LARGE LANGUAGE MODELS
Large Language Models (LLMs), such as OpenAI's
GPTs, have achieved a groundbreaking
advancement in the field of AI, particularly in
natural language processing. These models have
undergone extensive two-stage training: in the
pre-training stage, they are trained on a large
dataset without manual annotation, establishing
a foundational understanding of language; this is
followed by the fine-tuning stage, where the
models are trained on smaller, task-specific
datasets with manual annotations.

GENERATIVE AI
Generative artificial intelligence (GenAI) refers to
AI systems that can generate text, images, or
other media in response to prompts. GenAI are
based on LLMs and have an ability to generate
output that closely resembles human language.
Typically, users do not directly interact with
LLMs themselves but use specific GenAI tools
built upon them. For example, OpenAI's ChatGPT
is a conversational GenAI system fine-tuned for
dialogue tasks, based on GPT-3.5/GPT-4.
Similarly, Bard is a conversational GenAI chatbot
developed by Google, initially leveraging LaMDA
(Language Models for Dialog Applications) and
later incorporating PaLM (Pathways Language
Model).
FBE REPORT // JUNE 2023

06

GUIDELINE 1
UNDERSTAND GenAI

Ensure that you possess a comprehensive

understanding of the generative artificial
intelligence technology utilized in your legal
practice. Take the time to familiarize yourself
with its capabilities, functionalities, and
potential legal implications. This knowledge will
empower you to make informed decisions
regarding the appropriate and responsible
utilization of the technology, determining when
and how to leverage it effectively in your legal
work.

PAY ATTENTION TO:

It is essential to understand both the known
threats associated with machine learning, such as
bias and lack of interpretability, as well as the
risks specific to LLMs and GenAI, such as
hallucinations (making up information or content
that does not align with reality or factual
accuracy). Familiarize yourself with the main
principle of LLMs: their primary task remains the
prediction of the next token (simplified: the next
word in a string of words), with the result that
the answers generated by GenAI may or may not
be true.
FBE REPORT // JUNE 2023

07

GUIDELINE 2
ACKNOWLEDGE LIMITATIONS AND CONTEXT

Understanding how LLMs and GenAI can enhance

legal processes and offer more effective solutions
is crucial for remaining at the forefront of the
industry. However, it is important to consider the
limitations of this technology. Despite its
advancements, it inherently has constraints.
Acknowledge that AI-generated content may not
always be entirely accurate, comprehensive, or
up-to-date. Constantly bear in mind the primary
purpose of large language models, the risks of
hallucinations and bias. Remember that GenAI
systems that serve conversation, for example, are
not necessarily suitable for other applications,
including legal applications. Always verify the
output generated by the system.
FBE REPORT // JUNE 2023

08

GUIDELINE 3
ADHERE TO EXISTING RULES ON AI USE

It is essential for European lawyers to

acknowledge and adhere to the existing rules
governing the use of AI. This includes staying
informed about the EU Artificial Intelligence Act,
which is currently in the legislative process.
Additionally, it is crucial to be aware of any
possible national regulations or rules established
by Bar Associations, even if they are not legally
binding. Lawyers should also thoroughly review
the Terms of Service provided by GenAI providers
to ensure compliance with ethical and legal
standards.

It is important to note that the rules on AI-use are

subject to constant evolution and changes,
reflecting the dynamic nature of AI technology and
legal frameworks. Therefore, lawyers should
remain vigilant, stay updated on developments,
and adapt their practices accordingly to align with
the evolving regulatory landscape.
FBE REPORT // JUNE 2023

09

GUIDELINE 4
COMPLEMENT LEGAL EXPERTISE

GenAI should complement your legal

competences rather than replace them. Utilize it
as a tool to streamline processes in your law firm
and improve the efficiency of legal services.
However, remember that GenAI should not
substitute the exercise of your professional
judgment, legal analysis, and expertise.

Avoid becoming overly reliant on the outputs of

GenAI. Rememeber that while it can assist in
generating legal documents, it should not replace
human decision-making. Exercise caution and
independently verify the outputs through
analysis, research, fact-checking, and
professional judgment.

There is a risk that some individuals may opt not

to seek legal counsel because they perceive a
"consultation" with a chatbot based on LLMs as
sufficient. It is important to inform clients about
the risks associated with relying solely on this
type of practice.
FBE REPORT // JUNE 2023

10

GUIDELINE 5
MAINTAIN ATTORNEY-CLIENT PRIVILEGE

Safeguard attorney-client privilege when using

generative AI tools or platforms. Take necessary
precautions to protect the confidentiality of
client information and communications. Ensure
that the generative AI systems you use have
appropriate security measures in place.
Implement your own robust security measures to
safeguard sensitive data and adhere to ethical
and legal obligations related to client
confidentiality.

PAY ATTENTION TO:

Depending on the provisions of the Terms of
Service or equivalent regulations, entering
directly or indirectly client data as part of a
prompt in GenAI systems may not be permissible
due to the potential for further processing of the
information contained within the prompt.
FBE REPORT // JUNE 2023

11

GUIDELINE 6
ENSURE DATA PROTECTION AND PRIVACY

While partially overlapping with the previous guideline on client-

attorney privilege, this guideline underscores the crucial importance of
safeguarding personal data for lawyers when utilizing GenAI tools. It
emphasizes the need for a detailed assessment of GenAI's compliance
with the General Data Protection Regulation (GDPR), highlighting the
significance of exercising caution, implementing appropriate measures,
and addressing privacy concerns, and emerging risks. It needs to be
remembered that the professional use of GenAI introduces new threats
alongside the customary challenges of data protection and privacy.

Attorneys must be mindful that inputting personal data into genAI

systems requires a proper legal basis and assessment in compliance
with data protection and privacy provisions. Remember that GenAI tools
not only process data to generate responses to prompts but also
employ the provided data to enhance the system itself. However, these
risks can be partially mitigated by utilizing GenAI tools through
application programming interfaces (APIs) and/or by using a special
"opt-out" option, which can help separate the input data from system
development.

To ensure data privacy and protection, it is essential to implement

robust security measures encompassing both technology and processes,
guarding against unauthorized access, use, or disclosure of personal
data.
FBE REPORT // JUNE 2023

12

GUIDELINE 7
INFORM CLIENTS AND ASSUME RESPONSIBILITY

Maintain transparent communication with your

clients regarding the use of GenAI in your legal
practice. Clearly explain the fact that you use it,
as well as the purpose of such use, benefits,
limitations, and guarantees, ensuring that clients
understand the role of this technology in their
legal matters. Specify whether it is used for legal
research or additional office tasks.

Assume responsibility for the outcomes and

consequences of utilizing GenAI, while being
mindful of your professional obligations and
potential liability.

Keep in mind that it may be advisable to wait for

specialized systems designed to assist lawyers
(specifically tailored for the legal sphere) to
become available on the market. Avoid using
generic conversational generative AI systems that
may not be suitable for strictly legal applications
when automating substantive tasks in your work.

Adopt a responsible approach to using GenAI

tools within your law firm. Inform all your
coworkers about the established rules and ensure
their compliance.
FBE REPORT // JUNE 2023

13

SUMMARY
HOW LAWYERS SHOULD TAKE ADVANTAGE OF THE OPPORTUNITIES
OFFERED BY LARGE LANGUAGE MODELS AND GENERATIVE ARTIFICIAL
INTELLIGENCE

While we acknowledge the significant

advantages of using generative AI in the legal
field, we caution against hastily and
inappropriately applying GenAI tools to tasks
that lie at the core of legal competence and the
lawyer-client relationship.

We believe that by adhering to the presented guidelines, European

attorneys can effectively and responsibly integrate generative AI into
their legal practice. This will allow them to harness the technology's
benefits while upholding ethical standards, safeguarding client
confidentiality, and maintaining professional competence.

We strive to ensure that the ongoing business revolution driven by the

widespread use of large language models also benefits the professional
legal community. Simultaneously, we aim for lawyers within the Bar
Associations of FBE to be guided by the highest ethical standards in their
practice. These guidelines serve as a compass for lawyers who aspire to
embrace disruptive technologies while adhering to professional
requirements.
FBE REPORT // JUNE 2023

14

CALL TO ACTION
Shaping the Ethical Framework for LLMs and
GenAI in the Legal Sphere: A Call to Members of
the European Bars Federation

We encourage these guidelines to serve as an

invitation to initiate a comprehensive discussion
on the legal and ethical framework concerning
the use of GenAI within the legal sphere.

We kindly request you to consider these

recommendations and engage in deliberations
within your respective Bar Associations,
contemplating how they align with existing or
planned ethical rules and professional
standards. FBE is eager to know about the
outcome of these discussions.

It is important to acknowledge that the

guidelines may need to be subject to
modification over time, in response to
legislative changes or advancements in
technology itself. It is crucial to stay updated
and maintain a responsible approach towards
GenAI in the legal profession. This will establish
the foundation for a new era of delivering legal
services in Europe, where knowledge is kept
current, and a responsible approach to LLMs and
GenAI is embraced.