Yes, lawyers can use ChatGPT, but only with public law and text that holds no client data. Do not paste client names, identity numbers, bank details, whole files or anything covered by professional secrecy, and check every answer yourself.
Many lawyers and notaries already use ChatGPT to draft, summarize and explain. It saves time, but a chatbot is not a locked filing cabinet. Before you paste anything, you need to know where it goes and who can see it.
This guide is general information, not legal advice. Follow the guidance of your bar or chamber, and check it again from time to time, because it changes.
Can lawyers use ChatGPT?
In most places, yes. Nothing says you cannot use an AI chatbot. The question is what you give it. Your duty of professional secrecy and your duties under data protection law (GDPR) do not stop when you open a new browser tab.
So the real question is not "can I use ChatGPT?" but "what can I put into it, and in which kind of account?"
Which kind of account is safe for a law firm?
Not all accounts are the same, even for the same chatbot. In general terms, there are three kinds of setup:
- A free personal account. It is made for one person, not for an office. You have little control, and no contract that covers your clients' data.
- A business plan. These often come with more controls and a written contract. They are usually a better base for office use.
- A tool built for law firms. It is designed around documents, access rules and the duties of the profession.
Do not guess which one you have. Read the provider's settings and contract for these points:
- Is what you type used to train the model?
- How long is it kept, and can you delete it?
- Who at the provider, or at its partners, can see it?
- Is there a data processing agreement?
Plan terms change often, so check them yourself, today, and again later. A name like "pro" or "plus" does not tell you any of this.
What can a lawyer paste into ChatGPT?
Green: fine to use
- Public law: statutes, regulations, published decisions. Asking "explain this rule in simple words" is fine.
- Your own text that has nothing to do with a client: a blog post, a newsletter, a job ad, a talk.
- Generic templates and checklists with no real names, amounts or facts.
- General questions about how a type of contract works.
Orange: use with care
- Anonymized extracts of a real file. Anonymizing well is harder than it looks. Removing a name is not enough if the rest still points to one person: a rare job, a small village, an exact date, an unusual amount.
- Short pieces of a clause or a letter, with every detail that could identify someone removed or replaced.
- Facts of a case described in a general way ("a seller and a buyer of a house, with a dispute about a hidden defect").
Read the anonymized text again as a stranger would. If someone who knows the case could recognize it, it is not anonymized.
Red: do not paste
- Client names, addresses, phone numbers and emails.
- Identity numbers, tax numbers, bank details and passport or ID scans.
- Whole case files, deeds, contracts or court papers.
- Anything covered by professional secrecy, and anything a client told you in confidence.
- Medical, family or financial details about a person.
If you are not sure which colour something is, treat it as red.
Can you trust ChatGPT's answers?
A chatbot writes text that sounds right. It does not know when it is wrong. Two problems come up again and again:
- Made-up sources. A chatbot can give you a law, a court decision or a quote that looks real and does not exist, or that says something different from what it claims.
- Wrong details. Dates, deadlines, amounts and the name of the rule can be wrong, or taken from another country's law.
The fix is simple: never rely on an answer you cannot trace to a source. Open the source and read it yourself. A tool that shows where an answer comes from makes this much faster.
You stay responsible for what you sign and send. "The software said so" is not an answer to a client or a judge.
How do you choose an AI tool for a law firm?
Ask these questions, and ask for the answers in writing:
- Does it show its sources? Each answer should point to the document and page it comes from, so you can check it in seconds.
- Where is the data processed? Ask for the country and the company that runs the servers.
- Who are the subprocessors? These are the other companies that touch your data, such as the AI model provider, the email service and the error tracking. A serious provider names them.
- Can the office delete its data? You should be able to export your files and remove them when you leave.
- Is there a data processing agreement? Under GDPR, you need one with any provider that handles personal data for you.
- Is your data kept apart from other offices? Other firms must not be able to see your files.
- Is a person in the loop? Tools that hide or change data should let someone review the result before it is shared.
- What does the provider say about training? Get a clear statement on whether your data is used to improve models.
If a provider cannot answer these in plain words, keep looking.
Some simple habits for the whole office
- Agree on one rule for the team: what is green, orange and red. Put it on one page.
- Do not let staff use personal accounts for client work.
- Keep a short list of the tools the office has approved.
- When in doubt, ask before you paste. A paste cannot be taken back.
Where Acta fits
Acta is a web app for notaries, lawyers and law firms, built around the office's own documents. Ask Acta answers questions from those documents and shows the exact page each answer comes from, so you can check it. Anonymization hides personal data in a document before it is shared, and a person reviews what was removed.
Each office's data is private to its workspace, and members of other offices can't see it. Acta runs on a server in Germany and uses OpenAI's models for its AI features. Its privacy page names OpenAI, Resend (email) and Sentry (error reports) as subprocessors. The owner of an office can export everything or delete the workspace, which removes its files. Acta is free during the pilot.
If you would like to see how it works on your own kind of files, request a demo.
Frequently asked questions
- Can lawyers use ChatGPT?
- In most places, yes. Nothing says a lawyer cannot use an AI chatbot, but professional secrecy and data protection duties (GDPR) still apply. The real question is what you put into it and in which kind of account. Follow the guidance of your bar or chamber.
- What should a lawyer never paste into ChatGPT?
- Never paste client names, addresses, phone numbers or emails, identity numbers, tax numbers, bank details, ID scans, whole case files, deeds, contracts, court papers, anything covered by professional secrecy, or medical, family or financial details about a person.
- Is it safe to put client data into ChatGPT?
- Not by default. A free personal account is made for one person and has no contract covering client data. Check whether your input is used for training, how long it is kept, who can see it and whether there is a data processing agreement. When in doubt, keep client data out.
- Can ChatGPT make up legal sources?
- Yes. A chatbot can give a law, a court decision or a quote that looks real but does not exist, or that says something else. Dates, deadlines and amounts can also be wrong. Never rely on an answer you cannot trace to a source, and read the source yourself.
- How do I choose an AI tool for a law firm?
- Ask for written answers to these points. Does it show its sources? Where is the data processed? Who are the subprocessors? Can the office delete its data? Is there a data processing agreement? Is your data kept apart from other offices? What does the provider say about training?
Written by Ala Baganne, software engineer and founder of Acta.
About the authorSee what Acta can do for your office
Tell me which tasks take your office the most time, and I’ll tailor a free demo to them.
Request a demo