AI Training for Lawyers 2026: Program, Boundaries and Price (from €1,200 net per day)

AI training for lawyers starts with boundaries rather than prompts: what may be pasted into a model, in which environment, and who answers for the result. Only then does the team work on its own material: contract versions, case files, working correspondence. An engineer runs the closed workshop, from €1,200 net per day, for a group of up to about 15 people.

Quick answer

A legal team is an unusual audience for AI training, because every task touches two things at once: someone else's data and someone else's liability. So we build the workshop the other way round from a marketing or sales session. First we settle where acceptable use ends, then we teach the tool. We price it plainly, net:

• free process scan (€0): 30 minutes with an engineer plus a written takeaway in two business days, to establish what the team actually needs before you buy anything,
• closed AI training (from €1,200 net per day): a workshop on your documents, online or on-site, for one workshop group (up to about 15 people),
• at the end: a written way of working with AI, the rule for client data, and materials that slot into your AI Act article 4 competence documentation.

The full price list is on the pricing page, and the team offer is described on the AI training for teams page. If only one person needs training, one-to-one AI training is the better fit.

What a legal team actually works on

These four contexts come up most often when we scope a workshop. They are examples for that conversation. We do not present them as delivered law-firm engagements.

The left side of the table is the work a workshop genuinely speeds up. The right side is why training for lawyers spends more time on rules than on the tool.

The boundaries we settle first

Four decisions land before anyone opens a model.

Professional secrecy and client data. We build the list: what may be pasted, what may never be, and what needs pseudonymizing first. Plus one specific environment the team works in, with a data processing agreement on the vendor side and training on your data switched off. What to check with a vendor, we unpack in the analysis of Claude and GDPR.

Hallucination and the duty to verify. Models produce text that sounds equally confident when it is right and when it is invented. In legal work that trait is expensive, so the workshop rule is simple: every citation, quotation and legal basis goes back to the source before use.

The model does not decide the merits. The model prepares material, the lawyer rules on it. That line is not a matter of taste: professional responsibility cannot be moved onto a tool.

A trail of the work. It is worth knowing which document was produced with a model's help, who approved it, and which version they worked from. Five minutes of habit that pays for itself the first time anyone asks about the process.

The workshop program

Six modules, adapted to what the team really does.

1. How a model answers. Where the answer comes from, why it sounds confident when wrong, and which tasks are easy for a model and which are risky.
2. Context and task. How to describe a matter and what to ask for, so the output can be checked in reasonable time.
3. Work on the team's documents. Contract versions, case files, correspondence: the real tasks from the table above, done in the room and verified on the spot.
4. Data boundaries. The approved environment, pseudonymization, and the list of things that never get pasted.
5. Verification and escalation. The rhythm of checking sources, the point where the process stops, and the person who approves.
6. The written standard. The document that remains after the workshop and applies on Monday morning.

The last module is what separates training from an event. Without a written standard the team drifts back to old habits inside two weeks.

GDPR and the AI Act in a law firm

Three things need settling before the team works on client documents. First, a data processing agreement with the tool vendor, together with clarity on where data is physically processed; at Anthropic the DPA with standard contractual clauses is part of the commercial terms, and data from the API and business plans is not used to train models by default. Second, AI Act article 4, which since 2 February 2025 requires measures towards AI literacy among the people operating these systems; the workshop program, materials and attendance list are an input to that documentation. Third, article 50, whose transparency obligations apply from 2 August 2026 and are therefore already in force: if a client is dealing with an AI system and that is not obvious from context, they are entitled to know.

Data residency and access routes are covered separately in the piece on Claude and GDPR, and what article 4 actually requires in the guide to AI Act article 4 literacy training. This is not legal advice.

What AI training for lawyers costs

A closed workshop starts from €1,200 net per day. The price covers a day of work and one workshop group of up to about 15 people; we do not charge per head. So for a team of a dozen-odd lawyers it works out cheaper per head than sending each to a separate open course. The figure moves with the number of days, the number of groups (a larger team is split and priced in one offer) and the format: online or on-site. You get the final per-scope quote after one call.

Before you ask for an offer, work out what the document work you want to speed up costs you today. Fill in your own numbers:

Weekly cost of document work:
  hours per week per person × people on the team × hourly rate = weekly cost

Reference point for the workshop:
  workshop cost ÷ weekly cost = how many weeks of that work one training day costs

The result is a sense of scale rather than a promised return. If the weekly cost of that work is low, the workshop will be hard to justify, and it is better to know that before a sales conversation.

When NOT to book this training

• Only one person needs training. A group workshop makes sense when a shared standard has to emerge. For a single person, one-to-one training on their own process is cheaper and works better. We have delivered five such sessions; the team format is an offer.
• The partners have not agreed the data rules yet. With no decision on which tool is approved and what may be processed in it, the workshop gives the team practice at something the firm has not formally permitted. Set the company AI policy and employee rules first, then train.
• The goal is automation rather than competence. If the point is for documents to move through a process on their own, that belongs to a deployment conversation rather than a training room. Start with the process scan; if something simpler will do, we will say so.

How to start

1. Book a free process scan and tell us how many lawyers are meant to work with AI and on what documents.
2. Prepare three things: which tasks repeat every week, which tool the team already has, and whether a decision on client-data rules exists.
3. After the call you get a written recommendation: a closed workshop, one-to-one training for a single person, or an honest "policy first, training second".

Book a free process scan | AI training for teams | See pricing

FAQ

How much does AI training for lawyers cost?

A closed AI training for a legal team starts from €1,200 net per day. The price covers a day of work and one workshop group of up to about 15 people; we do not charge per head. The figure moves with the number of days, the number of groups and the format (online or on-site). You get the final per-scope quote after one call, and that call is free: 30 minutes with an engineer plus a written takeaway in two business days.

Is client data safe during the training?

The workshop starts by settling what may be pasted into a model and in which environment. We work in the tool your firm has approved, on documents the team itself clears, pseudonymized where needed. We make no absolute promises: safety depends on the vendor configuration, the data processing agreement and the rules the firm actually enforces. Training gives the team the rule and the habit; configuration closes the rest.

Can AI write court submissions?

A model can prepare a draft, organize the argument and point out places to check. Responsibility for the content stays with the lawyer and cannot be moved onto a tool. Every case citation, quotation and legal basis goes back to the source before use, because models can generate a judgment that does not exist. The workshop trains that verification rhythm on your own documents.

Online or on-site at our firm?

We run both. Online works well when the team is distributed or sits across several offices, while an on-site workshop wins where documents cannot leave the firm and we work on your machines and your environment. A workshop group holds up to about 15 people; a larger team is split into groups and priced in one offer.

Related articles

• AI training for teams
• Is Claude GDPR compliant
• AI Act article 4: the AI literacy obligation for companies
• Company AI policy: employee rules
• One-to-one AI training
• Syntalith pricing