P
Cas Pratiques

AI and Law: How to Use ChatGPT for Legal Work

By L'Art du PromptingPublished on June 9, 202614 min read

The average legal professional spends 30 to 40% of their time on repetitive tasks: reviewing clauses, checking compliance, summarizing case law. What if you could reclaim some of those hours every week?

ChatGPT won't replace your lawyer. Let's be clear about that. But used with the right prompts, it becomes a formidable legal assistant — capable of producing a first draft contract in 2 minutes, scanning a document for risks, or summarizing 50 pages of case law into one actionable paragraph.

The problem? Most people type "draft me a contract" and get a generic result riddled with approximations. Output quality depends entirely on prompt quality. And in law, approximations can be very costly.

In this article, you'll discover how to structure your prompts to get reliable, actionable legal results. With concrete examples, ready to copy-paste, adapted for practical use.

The legal sector is undergoing a quiet revolution. Legaltech companies are multiplying, law firms are integrating AI into their workflows, and even small businesses are starting to use ChatGPT for their daily legal needs.

Three factors explain this acceleration:

  • The volume of regulation is exploding. Between GDPR, industry-specific regulations, evolving employment law, and new digital acts, staying current has become a full-time job. AI lets you process this volume without burning out.
  • Legal costs remain high. A lawyer consultation starts at $300-500 per hour. For a startup or freelancer, using AI to prepare the groundwork before consulting a professional saves both time and money.
  • Model quality has crossed a threshold. GPT-4 and its successors understand legal nuances — provided you give them the right context.

Before diving into examples, you need to understand what separates a prompt that produces legal fluff from one that generates an actionable document.

ChatGPT was trained on law from around the world. If you don't specify your jurisdiction, you might end up with clauses from a completely different legal system — and that won't hold up in your local court.

Always mention: the country or state, the area of law (employment, commercial, consumer protection, etc.), and where relevant, the specific statutes or regulations that apply.

Rule #2: Provide the business context

"Draft a contract" isn't enough. The AI needs to know: who are the parties? What's the subject matter? What's the duration? Are there industry-specific considerations? The more context you provide, the more precise and usable the result.

Rule #3: Ask for sources and limitations

A good legal prompt always asks the AI to cite relevant statutes and flag areas of uncertainty. This is your safety net: you know exactly where to verify and where to have a professional review.

Rule #4: Never use as-is

AI produces a first draft. Not a final document. Every output must be reviewed, adapted to your specific case, and — for important documents — validated by a legal professional. This is non-negotiable.

If you're new to prompt engineering, the prompt builder can help you structure your requests with the right methodology.


Let's get practical. Here are tested, optimized prompts for the most common legal use cases. Each prompt is designed to produce a structured, actionable result.

1. Draft a service agreement

The service agreement is probably the most requested document by freelancers and small businesses. This prompt produces a solid foundation:

Terminal
You are a lawyer specializing in commercial law. Draft a service agreement between: - Service Provider: [Company name], [legal form], registration number [number], headquartered at [address] - Client: [Name], [legal form], registration number [number], headquartered at [address] Subject of the service: [describe the service precisely] Duration: [planned duration] Amount: [amount] excluding taxes Payment terms: [net 30, in 3 installments, etc.] Must include: - Intellectual property clause (rights transfer to client upon full payment) - Confidentiality clause - Limitation of liability clause - Termination conditions (with 30-day notice) - Force majeure clause - Jurisdiction clause ([city/state] courts) - Data protection compliance if personal data is processed Cite the relevant statutes for each clause. Flag any points that require review by a licensed attorney.

This prompt works because it provides the full context: parties, subject, amount, and the specific clauses expected. The AI doesn't have to guess — it executes.

2. Analyze GDPR / data privacy compliance

Data privacy remains a major concern for businesses. This prompt turns ChatGPT into a first-level privacy auditor:

Terminal
You are a Data Protection Officer (DPO) with expertise in GDPR and data privacy law. Analyze the compliance of the following data processing activity: - Data controller: [company name] - Processing purpose: [e.g., sending marketing newsletters] - Data collected: [e.g., email, first name, purchase history] - Legal basis: [e.g., consent / legitimate interest / contractual necessity] - Retention period: [e.g., 3 years after last contact] - Processors: [e.g., Mailchimp (USA), hosting on AWS (EU)] - Transfers outside EU/EEA: [yes/no, to which country] For each point, assess: 1. Compliance with GDPR (Articles 5 to 49) 2. Identified risks (low / medium / high) 3. Recommended corrective actions 4. References to relevant regulatory guidance Present the result as a table with columns: Item checked | Status | Risk | Corrective action | Legal reference

This type of automated audit doesn't replace a real DPO, but it lets you identify obvious gaps before paying for a consultation. You'll find more prompts like this in our prompt library.

3. Generate compliant Terms and Conditions

Terms and Conditions are mandatory for any e-commerce business. Poorly drafted, they expose you to legal liability. This prompt produces structured T&Cs:

Terminal
You are a lawyer specializing in consumer protection law. Draft comprehensive Terms and Conditions for: - Business: [name], website [URL], activity: [online sale of physical products / digital services / SaaS] - Target: consumers (B2C) The T&Cs must cover: 1. Purpose and scope 2. Pricing and payment terms 3. Right of withdrawal / cancellation policy (14-day cooling-off period where applicable) 4. Delivery and risk transfer 5. Legal warranty of conformity and warranty against hidden defects 6. Liability 7. Personal data protection (GDPR) 8. Dispute resolution and mediation 9. Governing law and jurisdiction For each section, cite the applicable legal provisions in parentheses. Use clear, accessible language in compliance with pre-contractual information obligations. Flag any provisions that must be adapted based on the specific business activity.

4. Draft a formal demand letter

A formal demand letter is often the first step before legal action. It must be firm, precise, and legally grounded:

Terminal
You are a lawyer specializing in contract law. Draft a formal demand letter with the following elements: - Sender: [name, full address] - Recipient: [name, full address] - Subject of dispute: [e.g., unpaid invoice #XXX dated MM/DD/YYYY for $XXX] - Context: [summarize the facts: service performed on..., invoice sent on..., reminders sent on...] - Legal basis: breach of contract under applicable contract law - Deadline: 15 days from receipt - Consequences of non-compliance: filing with the competent court, application of late payment penalties as provided by law Tone: firm and professional, not aggressive. Format: formal letter with header, date, references, body, and closing. Mention sending via certified mail with return receipt.

The demand letter is a document where the precision of facts and legal references makes all the difference. Note how the prompt specifies tone and format — two elements AI handles very well when made explicit.

5. Summarize and analyze a court decision

Legal research is time-consuming. This prompt saves you hours on case law analysis:

Terminal
You are a legal researcher specializing in case law analysis. Analyze the following court decision: [Paste the full text of the decision, or indicate: Court, division, date, case number] Structure your analysis as follows: 1. CASE BRIEF - Court and date - Parties (plaintiff / defendant) - Relevant facts (chronology) - Procedural history - Legal question at issue - Court's ruling 2. ANALYSIS - Legal rule applied (statutes, principles) - Court's reasoning - Scope of the decision (reversal? confirmation? clarification?) - Practical impact for practitioners 3. BROADER CONTEXT - Prior case law on the same issue - Evolution of court positions - Concrete consequences for practice Use a clear, concise style. Maximum 800 words.

6. Audit contract clauses

Received a contract to sign and want to spot the traps before committing? This prompt is made for that:

Terminal
You are a business lawyer specializing in contract law. Analyze the following clauses extracted from a [contract type: service / license / partnership] agreement: [Paste the clauses to analyze] For each clause, assess: 1. Legal validity under applicable law 2. Balance between the parties (unconscionable clause? significant imbalance?) 3. Risks for [specify: the provider / the client] 4. Recommended rewording if the clause is imbalanced or risky 5. Missing clauses that should be added Rate each clause by risk level: Acceptable | Needs modification | Dangerous
Terminal
You are a lawyer specializing in digital law. Draft the mandatory legal notices for the following website: - Publisher: [company name, legal form, registration number, headquarters address] - Publication director: [name] - Host: [name, address, phone] - Activity: [describe the website's activity] - Tax ID: [if applicable] In compliance with applicable regulations for website operators, including data protection laws. Include sections for: publisher identification, hosting provider, intellectual property, personal data protection (with DPO mention if applicable), cookies, credits. Flag any specific mandatory disclosures based on the business sector.

8. Analyze an employment law issue

Terminal
You are a lawyer specializing in employment law. Analyze the following situation: [Describe the situation: e.g., An employee on a permanent contract for 3 years wants to challenge their dismissal for gross misconduct. The alleged facts are... The procedure followed was...] In your analysis: 1. Identify the applicable statutes and regulations 2. Assess the legal characterization of the facts 3. Verify the regularity of the procedure 4. Estimate the risks for the employer / the employee 5. Propose available remedies with applicable limitation periods 6. Estimate potential compensation based on applicable scales and precedent Present a structured summary with the strengths and weaknesses of each party's position.

The prompts above cover common needs. But legal AI goes much further when you know how to push the boundaries.

Instead of scanning legal publications every morning, you can ask AI to synthesize the developments that matter to you. Combine ChatGPT with an RSS feed of legislative publications, and use a prompt like:

"Analyze the following texts published this week. Identify those that impact [your industry]. For each relevant text: 3-line summary, concrete impact, actions to take, deadline if any."

Our ready-to-use AI agents can fully automate this type of workflow — an agent monitors, summarizes, and alerts you only when it's relevant.

Comprehensive compliance audit

You can build a chained prompt that sequentially audits: data privacy, legal notices, terms and conditions, cookie policy, and subprocessor agreements. Each step feeds into the next. This is the principle of "prompt chaining" — a technique you can master with our practical exercises.

Intellectual property and AI

A hot topic: who owns the rights to AI-generated text? In most jurisdictions, copyright protects "works of authorship" involving original human creative contribution. A purely AI-generated text, without creative human intervention, is likely not protectable.

Practically speaking, if you use ChatGPT to draft a contract or terms and conditions, you should rework the text enough to add your own personal touch. Paradoxically, AI forces you to be more creative, not less.

Compliance and commercial law

For businesses subject to compliance obligations (anti-corruption laws, duty of care, international sanctions), AI excels at screening. Give it your risk mapping and ask it to identify gaps. It's not a substitute for a structured compliance program, but it's a powerful accelerator for first-level analysis.

Every legal situation is unique. Rather than searching for the perfect prompt each time, learn to build your own. The prompt builder guides you step by step: you select the legal area, document type, and desired level of detail, and you get a structured prompt tailored to your exact need.


Limitations to know (and respect)

Enthusiasm shouldn't overshadow the risks. Here's what AI cannot do in legal matters:

  • Provide personalized legal advice. AI provides general information. Only a licensed attorney can give you legal advice that carries professional liability.
  • Guarantee 100% accuracy. Models can "hallucinate" — inventing statutes that don't exist or citing fictional case law. Always verify against official sources.
  • Replace recent case law. Models have a knowledge cutoff date. A ruling from last week won't be in the knowledge base.
  • Handle confidentiality. Never paste personal data or confidential documents into ChatGPT without checking the service's privacy policy and, ideally, using the API or Enterprise version with training opt-out.

FAQ: AI and law in practice

Can ChatGPT replace a lawyer?

No. ChatGPT is a drafting and analysis tool, not a legal professional. Unauthorized practice of law is a serious offense in most jurisdictions. AI can prepare a first draft, identify risks, and summarize case law — but final legal advice and court representation remain the domain of licensed attorneys. Use AI to arrive better prepared at your lawyer's office, not to avoid going there.

A contract or legal document is legally valid as long as it meets the standard conditions of validity (mutual consent, capacity, lawful purpose). It doesn't matter whether it was drafted by a human, an AI, or a medieval scribe. What counts is the content and the parties' signatures. However, a poorly drafted AI document could contain void or unenforceable clauses — which is why human review is essential.

Three techniques that work: (1) always ask the AI to cite exact statutes and verify them against official legal databases; (2) use the prompt "If you're uncertain about a legal point, explicitly state so rather than making something up"; (3) cross-reference results with a second source (legal databases or a second AI model). Hallucinations are more common with case citations (fabricated cases) than with statutes — be especially vigilant about case references.

Can I use ChatGPT for my company's GDPR compliance?

Yes, but as a pre-audit tool, not a complete solution. ChatGPT can help you map your data processing activities, identify appropriate legal bases, draft privacy clauses, and prepare your processing records. But GDPR compliance is an ongoing process requiring intimate knowledge of your business. For high-risk processing (sensitive data, profiling, transfers outside the EU), have it validated by a qualified DPO.

Am I taking a risk by using AI to draft my contracts?

The risk isn't in using AI, but in using it without verification. If you sign a contract containing an unconscionable clause generated by ChatGPT, you bear the consequences — not OpenAI. The real danger is false confidence: believing that because the text "looks legal," it's legally sound. Treat every AI output as a first draft, not a final document. And for high-stakes matters (fundraising, employment contracts, litigation), invest in a lawyer.

GPT-4 and newer models are significantly better than GPT-3.5 for legal tasks. They understand nuances better, structure documents more effectively, and hallucinate less. If you work regularly on legal documents, a ChatGPT Plus subscription or API access is worth the investment. Anthropic's Claude models are also strong for legal analysis, with a larger context window — useful for analyzing long contracts.


Take action

You now have a complete toolkit for using AI in your daily legal work. The prompts in this article cover 80% of common needs: contracts, data privacy compliance, terms and conditions, demand letters, legal monitoring, and case law analysis.

But the real power is knowing how to build your own prompts tailored to your situation. Every case is different, every clause has its subtleties.

To go further:

AI doesn't replace the law. It makes the law more accessible. And those who master this tool today will have a considerable competitive advantage tomorrow — whether you're a lawyer, entrepreneur, or simply someone who wants to understand their rights.

L'Art du Prompting

L'Art du Prompting

Founder of Prompt Guide and CEO of Webpulser. Digital and AI entrepreneur since 2006, he shares his field-tested prompt engineering techniques.

Stay Updated

Get our best articles and techniques every week.

Related Prompts

✍️RedactionIntermediateAll AIs

Transform long content into catchy social media posts

Transform your articles and long-form content into short, engaging social media publications.

301406
🎨CreativiteBeginnerAll AIs

Create Catchy Social Media Headlines

Generates punchy, optimized headlines for your social media posts, adapted to your audience and platform.

298356
✍️RedactionIntermediateAll AIs

Summarize long articles or documents into clear summaries

Condense articles, reports, and lengthy documents into structured and directly usable summaries.

297321
✍️RedactionAdvancedAll AIs

Write Professional Blog Posts and White Papers

Create engaging blog posts and professional white papers that strengthen your industry expertise.

296317

Also explore