AI for law firms: the case file never leaves the firm
- Updated
- By
- The MagnolIA team
- On this page
- 12 min read
Short answer
A law firm can use AI without handing its case files to a third party. MagnolIA installs a machine in the office, and lawyers and paralegals use it to find clauses, review documents and draft from contracts, case files and case law. Documents, questions and answers are processed and kept inside the firm.
What happens to professional secrecy when a case file is uploaded to a public AI?
Nearly everything that crosses a lawyer's desk is covered by professional secrecy, the Spanish counterpart of legal professional privilege and the duty of confidentiality. Spain's General Statute of the Legal Profession (Royal Decree 135/2021) states in article 22 that it covers all facts, communications, data, information, documents and proposals the lawyer has learned, issued or received in practice. The same article requires lawyers to make their associates and staff respect it, and the duty does not end when the engagement does.
When someone pastes a claim or a contract into a cloud AI, the document travels inside the prompt and is processed on another company's servers. An enterprise plan sets out by contract how it is handled, but it does not change the route: the file has left the firm.
Bar associations are already addressing it. In Spain every practising lawyer belongs to a local bar (a colegio), which sets and enforces professional conduct rules. The Madrid Bar Association (ICAM) presented its good practice guide on the use of AI in the legal profession in October 2025. According to its launch note, it asks firms to guarantee data protection and confidentiality, to hold third-party tools to the same standards as internal ones, and to appoint people responsible for oversight who ensure human review.
What does AI do in a law firm's daily work?
The reference deployment for a law firm is described in the on-premise AI use cases. It is used by lawyers, paralegals and administrative staff.
- Find clauses and review documents. Open the contract or the file as a PDF or Word document and ask it: deadlines, penalties, parties, what is missing.
- Draft briefs and opinions. The AI drafts from the documents of the matter; the lawyer reviews, corrects and signs.
- Query the firm's own documents. The server supports semantic search and RAG over what the firm holds.
- Get the minutes of a meeting. MagnolIA Nexus transcribes Meet, Zoom or Teams meetings with the speech model on your server and leaves minutes with a summary. The audio does not leave.
- Anonymise a document before sharing it. The Anonymiser agent in Nexus replaces names, ID numbers, bank accounts and other data, and restores them afterwards with its key.
- Hand errands to Magni. The personal assistant in Nexus works with the authorised folders, within the permissions granted.
What goes in and what comes out of each task?
| Task | Goes in | Comes out |
|---|---|---|
| Reviewing a contract | The contract as a PDF or Word file | Clauses located and points to watch |
| Preparing a brief | The documents of the case file | A Word draft to review |
| Checking precedents | The documents and case law the firm holds | An answer based on those documents |
| Client meeting | The audio of the meeting | Transcript and minutes with a summary |
| Sending to a third party | A document with personal data | A copy without personal data and the key to revert it |
What about working with third parties? Reversible anonymisation
Some documents do have to leave: to an expert witness, to another firm, to an external tool. The Nexus anonymiser flags each piece of personal data with rules and with a model running on your server; you review, and out comes a copy in which every item is a label. It detects 18 types of data, from Spanish ID numbers (DNI) and IBANs to names, addresses and job titles.
The mapping table stays on your computer: with it, whatever comes back has the real data restored; without it, nobody can undo the substitution.
In GDPR terms this is pseudonymisation (article 4(5)): it lowers risk and does not replace the assessment of each processing activity. The detail is in document anonymisation with AI, and you can try it in the online document anonymiser.
What safeguards does the European legal profession ask for before client data goes into an AI?
The Council of Bars and Law Societies of Europe (CCBE) published its guide on the use of generative AI by lawyers on 2 October 2025. Its basic rule (section 4.1) is to refrain from entering any personal, confidential or other client-related data into a generative AI tool unless appropriate safeguards are in place. It gives three examples, and none excludes the others.
| Safeguard (CCBE example) | What the firm has to do | What still needs watching |
|---|---|---|
| Contractual duty of confidentiality or zero data retention | Read the provider's terms and buy the plan that puts it in writing. | The case file still leaves the firm; the protection is the contract's. |
| A data protection agreement limiting use to the firm's own purposes | Sign it before using the tool on live matters. | Which other companies are involved, and the country where data is processed. |
| A system that runs locally or in a secured environment controlled by the firm | Have the machine on the firm's network and someone to keep it running. | Internal permissions, backups and who can get into the machine. |
The guide adds two warnings that apply whichever tool is chosen. First, analyse the provider's terms and conditions to understand how the data entered is used. Second, generative AI is now built into translation tools, PDF readers and text editors, nearly always cloud-based, and the same care is needed there. The general questions for any provider are in what to ask before giving confidential documents to an AI.
What if the AI gets it wrong or invents a citation? Who is responsible?
Using AI does not shift responsibility onto the tool: whoever signs the document still answers for it. A language model can present a judgment that does not exist, or credit a clause with something it does not say, in exactly the tone it uses when it is right. Keeping the AI inside the firm settles where the case file travels; it does not make the model more accurate.
The CCBE guide treats this as part of professional competence (section 4.2): lawyers are responsible for their work and for the representations they make to courts, and should verify the output of a generative AI before using it where the use case requires. It warns that using AI-generated content without appropriate verification could lead to sanctions for professional misconduct or to malpractice claims.
- How is a citation checked?
- By opening the decision in the firm's legal database or in the original document. If the AI does not show which document a statement comes from, that statement is not used until it has been found.
- Does the client have to be told?
- The CCBE guide (section 4.4) recommends transparency where it can reasonably be assumed that an informed client would object to, or set conditions on, that use. What goes in the engagement letter is for the firm to decide under its own bar's rules.
- And inside the firm?
- The guide notes the conflict-of-interest risk when one system has access to confidential information from several clients (section 4.5). On the firm's own machine that is handled with individual users and role-based permissions.
- Who oversees it?
- The Madrid Bar asks firms to appoint people responsible for oversight who ensure human review. In a small firm it can be a single named person.
What should AI not do in a law firm?
It is a tool for reading and drafting. These are the tasks best left out, or always done with a person in the loop:
- Filing or sending a document that its signatory has not read. The draft saves time; the review is not delegated.
- Citing case law or legislation nobody has opened. Every citation is checked at source before it leaves the firm.
- Being the only source for a deadline. A procedural time limit is checked against the diary and the rule itself.
- Deciding case strategy. It can organise the documents and point out weak spots; the judgement is the lawyer's.
- Replying to a client unsupervised. Whatever reaches the client carries the firm's name.
- Being used from personal accounts on public services. That use falls outside all three safeguards above.
Where does a law firm that wants to use AI start?
Find out what is already in use
Ask, without blame, which AI tools each person uses today, including those built into the translator or the PDF reader.
Decide what information can go where
Separate what is public (legislation, templates with no data) from what is covered by secrecy. The second kind needs one of the safeguards above.
Name the person who oversees it
Someone who approves tools, answers questions and reviews how they are being used.
Write short rules
One page on what may be entered, what is always reviewed and whom to ask. There is a starting point in the house rules for AI in the office.
Pilot with one type of matter
Reviewing lease agreements, for example. A narrow pilot shows where the AI gets it right and where it needs correcting.
Checklist before using it on a live matter
- Do we know where the document is processed and who outside the firm could see it?
- Have we read the terms on retention and training, or is the machine on our own network?
- Does each person sign in as themselves and see only their own matters?
- Is there a record of who asked what?
- Is it clear who reviews each draft before it goes out?
- Is every citation checked at source?
- Have we decided what we tell clients about our use of AI?
- Is there a named person responsible for overseeing use?
How is it installed, and what stays inside the firm's network?
MagnolIA delivers it installed and keeps it running: the firm does not administer servers or models. The reference deployment starts on Núcleo S, the compact node, and capacity is added without interrupting service if the firm grows.
Assessment
We look at how many people will use the AI, what for, and which systems it needs to talk to. That sets the capacity.
Installation
MagnolIA installs the machine in your office. You do not buy it: it is included in the fee.
Tuning and integration
The models are loaded and tuned, and the integrations with your systems are set up.
Pilot
A group of 10 to 20 people works with it on their own documents.
Production
The whole staff, with corporate sign-in, monitoring and backups.
- Stays inside
- Prompts, documents and answers. They are processed on the machine installed in your network and stay there.
- Also inside
- The per-user usage log and a chained, verifiable audit trail, kept for two years.
- Leaves
- Only the monitoring of the machine's status: load, health and security. MagnolIA does not see the content.
- Access
- Each person signs in with their own user and password through a single entry point encrypted with TLS. Remote access is optional and runs through an authenticated outbound tunnel.
Nobody sees a colleague's conversations: permissions are assigned by role. The measures are detailed in on-premise AI security, the payment model in what on-premise AI costs and the basic definition in what on-premise AI is.
Can you try it first? Review a contract in the LAB
Yes, for free. In AI contract review you upload a contract and receive a one-page brief: parties, deadlines, amounts, penalties and clauses to watch. The model only sees the anonymised text. You can also ask questions about an anonymised document.
Frequently asked questions from law firms
Can a law firm use AI without breaching professional secrecy?
A law firm can use AI without handing the case file to a third party. With a MagnolIA machine in the office, lawyers and paralegals find clauses, review documents and draft briefs and opinions, and everything is processed inside the firm's network. How professional secrecy applies to each tool is for the firm itself to assess.
What happens if AI invents a citation in a legal document?
If AI invents a citation in a legal document, responsibility does not pass to the tool. The October 2025 guide from the Council of Bars and Law Societies of Europe (CCBE) notes that lawyers are responsible for their work and recommends verifying AI output before it is used: every citation is opened at its source.
What safeguards does the European legal profession ask for when using AI with client data?
The October 2025 guide from the Council of Bars and Law Societies of Europe (CCBE) recommends not entering client data into generative AI without appropriate safeguards. It gives three examples: contractual duties of confidentiality or zero data retention, a data protection agreement, or a system running locally or in a secured environment controlled by the firm.
Does a law firm have to tell clients it uses AI?
Whether a law firm tells its clients that it uses AI is for the firm to decide under its own bar's rules. The October 2025 guide from the Council of Bars and Law Societies of Europe (CCBE) recommends transparency where it can reasonably be assumed that an informed client would object to, or set conditions on, that use.
Is anonymising documents enough to comply with the GDPR?
Anonymising is not enough on its own. What MagnolIA Nexus does is pseudonymise (Article 4(5) GDPR): it replaces the data and keeps the information needed to restore it separately, on your machine. It is a technical measure cited in Article 32 and applies data minimisation, but it does not replace assessing each processing activity.
Can an anonymised document be reversed?
A document anonymised with MagnolIA Nexus can be reversed with its key. Each replacement is stored as a mapping on the user's machine, and whatever the AI produces comes back with the real data. That key never leaves your machine: without it nobody can undo the anonymisation. Within a project, the same value always gets the same placeholder.
Sources
External figures on this page link to their original source.
BOE (Spain's Official State Gazette). Royal Decree 135/2021 approving the General Statute of the Spanish Legal Profession, articles 21 and 22 (in Spanish). www.boe.es/buscar/act.php?id=BOE-A-2021-4568
ICAM (Madrid Bar Association). Good practice guide on the use of artificial intelligence in the legal profession: launch note (in Spanish). 7 October 2025. web.icam.es/el-icam-lanza-la-primera-guia-practica-para-un-uso-responsable-de-la-ia-en-la-abogacia/
Council of Bars and Law Societies of Europe (CCBE). CCBE guide on the use of generative AI by lawyers (PDF). 2 October 2025. www.ccbe.eu/fileadmin/speciality_distribution/public/documents/IT_LAW/ITL_Guides_recommendations/EN_ITL_20251002_CCBE-guide-on-the-use-of-the-use-of-generative-AI-for-lawyers.pdf
EUR-Lex. Regulation (EU) 2016/679, General Data Protection Regulation (GDPR). eur-lex.europa.eu/eli/reg/2016/679/oj
Shall we look at it with your firm's documents?
Tell us how many people would use it and on what kind of matters. The pilot runs with a group of 10 to 20 people.
Request a pilot


