AI in Legal Advisory: Opportunities, Limitations and Ethical Questions

Artificial Intelligence Enters the Law Firms

Legal work is facing a paradigm shift. Artificial intelligence (AI) is no longer a topic for the future but is increasingly being adopted by law firms and legal departments. Large language models such as GPT-4, Claude or Gemini can analyse, summarise and even draft legal texts. At the same time, these technologies raise fundamental questions: Where are the limitations? Which professional obligations apply? And how can you, as a lawyer, deploy AI responsibly without putting your clients at risk?

This article provides a comprehensive overview of the current state of AI in legal advisory and outlines practical approaches for meaningful integration into your daily practice.

Areas of Application: Where AI Already Adds Value in Legal Advisory

Contract Analysis and Due Diligence

One of the most mature areas of application is automated contract analysis. AI-powered tools can review hundreds of contracts in a short time, performing tasks such as:

Identifying risk clauses (e.g., change-of-control clauses, liability limitations, termination rights)
Flagging deviations from standard contracts
Detecting missing clauses and highlighting industry-standard provisions
Extracting key terms (durations, notice periods, volumes)

In M&A transactions, this can significantly accelerate the due diligence process. Where teams previously spent weeks reviewing data rooms, AI tools can complete an initial review in hours and present results prioritised by relevance.

Legal Research

Traditional case law research also benefits from AI. Modern legal research tools enable:

Semantic search rather than pure keyword search – the AI understands the legal context
Summaries of court decisions in a few sentences
Tracing lines of argumentation across multiple decisions
Automatic identification and linking of relevant literature

Document Automation

AI can significantly reduce repetitive drafting work:

Drafting contracts based on templates and client requirements
Structuring legal briefs in preliminary form
Generating standard correspondence automatically
Preparing client updates (legal updates, newsletters)

Further Areas of Application

Compliance monitoring: Automated tracking of regulatory changes
Deadline management: Intelligent reminder systems with contextual understanding
Billing: Automated time tracking and billing optimisation
Conflict checking: Automated screening for conflicts of interest

The Hallucination Problem

The most serious weakness of current AI systems is so-called hallucinations. Large language models generate statistically plausible text – but they have no genuine understanding of legal accuracy. In concrete terms, this means:

Fabricated judgments: AI can generate case references, dates and even headnotes of court decisions that never existed
Incorrect statutory references: Paragraphs are correctly formatted but substantively misattributed
Outdated legal position: Training data is not up to date, so legislative changes may not be reflected
Logical fallacies: Legal arguments can appear superficially coherent but be doctrinally flawed

In the United States, the case of Mata v. Avianca attracted attention when a lawyer incorporated fictitious judgments generated by ChatGPT into a legal brief. Such cases underscore: every AI output must be verified by a qualified legal professional without exception.

Contextual Understanding and Case-by-Case Adjudication

AI frequently cannot grasp the nuances of a specific case. German jurisprudence is characterised by case-by-case analysis, the weighing of competing interests and systematic interpretation. This profound legal judgment – the so-called "legal subsumption" – remains a genuinely human achievement.

Data Quality and Training Bias

AI models are only as good as their training data. In the German legal sphere, particular challenges arise:

Limited German-language legal training data compared to the English-speaking world
Restricted accessibility of many relevant databases (juris, Beck-Online) for AI training
Systematic biases can lead to one-sided results

Professional and Ethical Obligations

Requirements under the BRAO

The Bundesrechtsanwaltsordnung (BRAO) sets clear requirements for legal practice that become particularly relevant when deploying AI:

§ 43a BRAO – Conscientiousness: The duty to practise conscientiously requires that you always critically review AI outputs. Responsibility for the correctness of legal advice remains exclusively with the lawyer.
§ 43a Abs. 2 BRAO – Confidentiality: Entering client-related data into cloud-based AI tools may violate attorney-client privilege. Always verify where data is processed and stored.
§ 43e BRAO – Use of service providers: When using external AI services, the provisions governing the engagement of service providers must be observed.

Data Protection Requirements

The use of AI tools in the law firm is subject to the strict requirements of the GDPR:

Legal basis: Processing personal client data through AI providers requires a sound legal basis
Data processing agreement: A data processing agreement (DPA) pursuant to Art. 28 GDPR must generally be concluded with AI providers
Third-country transfers: For US providers, the permissibility of data transfers to third countries must be assessed
Transparency: Clients should be informed about the use of AI tools

Ethical Guidelines

The bar associations and the Deutscher Anwaltverein (DAV) are increasingly developing guidelines for AI use. Core principles include:

Human final decision-making on all client-related matters
Transparency towards clients regarding AI use
Regular review of deployed tools for reliability
Continuing education obligation regarding technological competencies

The German Legal Tech Landscape

The German legal tech market is developing dynamically. Key areas where German startups and providers are delivering innovative solutions include:

Contract management: Platforms for automated contract creation and administration
Legal rights automation: Tools that automatically help consumers enforce their rights (e.g., flight passenger rights, tenancy law)
Practice management: AI-powered workflow optimisation and case management
Compliance solutions: Automated regulatory monitoring and reporting

The Bundesrechtsanwaltskammer is monitoring these developments closely and progressively adapting the regulatory framework.

Practical Implementation Tips for Law Firms

Phased Introduction

Do not start with the most complex use case; instead, proceed step by step:

Define a pilot project: Start with a clearly defined area of application (e.g., research support)
Evaluate suitable tools: Compare providers with regard to data protection, accuracy and costs
Create internal guidelines: Establish which data may be entered into AI tools and which may not
Conduct training: All staff must learn how to use AI tools safely
Establish quality control: Implement a four-eyes review process for AI-generated content

Specific Security Measures

Anonymisation: Remove or pseudonymise personal data before entering it into AI tools
Consider on-premise solutions: For sensitive areas, consider local AI models
Access restrictions: Not all staff need access to all AI functions
Logging: Document AI use for quality assurance purposes

Conclusion: AI as a Tool, Not a Replacement

Artificial intelligence will not replace legal advisory, but it will fundamentally change the way legal services are delivered. The key lies in responsible use: leveraging AI as a powerful tool without surrendering human expertise and legal judgment.

Law firms that engage with AI early and strategically will gain a competitive advantage – through greater efficiency, better quality control and innovative client services. Professional obligations and the protection of client interests must never be lost from sight.

compleneo supports you in identifying the potential of AI in your practice and implementing it in a legally compliant manner. From strategy development through tool selection to implementation, we are at your side as a competent partner.