AI policy, GDPR and the AI Act
A workable policy, the AI literacy the regulation asks of organisations, and the GDPR questions that come with every tool.
AI policy, GDPR and the AI Act is a half-day session in which the team writes a workable AI policy: which tools may be used, which data may not, who decides and how the team is trained. The AI Act asks organisations to take measures that support AI literacy. The module turns that duty into a register, training per role and a policy page the team actually uses.
- 01The team builds a register of the AI tools already in use, with owner and data type per tool
- 02The team writes a one-page policy with permitted tools, prohibited data and a decision-maker
- 03The team decides per role which training is needed to evidence AI literacy
- 04The team tests each new use for whether the GDPR or a high-risk rule of the AI Act applies
01
What the AI Act asks of you
Article 4 asks organisations to take measures that support AI literacy. What that means per role, and why the regulation does not prescribe one fixed course.
02
The register
Which AI tools already exist, who uses them, which data goes in, and who owns them. The list is often longer than expected.
03
The policy on one page
Allowed, allowed under conditions, and forbidden. With the rule on personal data, and the question who may approve a new tool.
04
Training and check per role
Which role needs which knowledge, and how you record it. Then the test per new use: GDPR and high risk.
What the AI Act asks of organisations
The AI regulation (Regulation (EU) 2024/1689) has been partly in force since 2025 and was amended in 2026 through the digital omnibus. Article 4 concerned AI literacy. According to the amended text in Regulation (EU) 2026/1744, providers and deployers of AI systems take measures to support the AI literacy of their staff and of others who work with AI systems on their behalf. The duty does not guarantee a set level of literacy for every individual. That is a real shift: it concerns measures and evidenced effort, not a result each person must reach.
For the team that means: you must be able to show what you do to keep your staff’s knowledge at the right level. The form is yours to choose. A register, training per role and a policy that is kept current together form the evidence. What the supervisors in the Netherlands expect is set out in the guidance the Dutch data protection authority has published on AI literacy. We take that into the module.
The register: what is already running
The first work is an inventory. Which AI tools does the team use, with or without a business account? Who uses them, for which work? Which data goes in: email text, client data, documents, speech? Who owns the account and the settings? The list is almost always longer than expected, because staff use tools nobody has approved. That is not a scandal, but it is the starting point of the policy.
In the module we build the register together, with a column per tool for the owner and a column for the type of data. The register is a working document. It is updated when something is added, and it is the basis for every later check.
The policy on one page
A policy nobody reads does not help. It must fit on one page and say three things. Which tools are allowed, and under which conditions. Which data never belongs in an AI tool, for example personal data of clients without a business account and a data processing agreement, and negotiation files. And who may approve a new tool, and within which period.
The text of the policy is a choice, not a legal template. We write it in the team’s own language, with examples from their own work. That makes it usable, and it is the reason staff actually read it.
The GDPR: the questions that come with every tool
The GDPR continues to apply alongside the AI regulation. For every AI application that processes personal data, the GDPR asks for a legal basis, a data processing agreement with the vendor, an assessment of transfers outside the EU, and, for a large or sensitive processing, a data protection impact assessment, a DPIA. The module walks through those questions for the tools in the register.
Mind the vendor. Some business versions promise that data is not used to train models; others do not, or only on request. That lives in the terms and the processing agreement, not in the marketing. We show where the team finds those documents and what they must contain.
High risk, and the check for each new use
The AI Act sorts systems by risk. Most business uses fall under limited or minimal risk. A system that decides about people, for example in recruitment or credit assessment, can fall under the high-risk rules. The date on which those rules apply has moved to later dates in 2026; check the current date before you decide.
The module ends with a check the team uses for each new use: which data, which decision, which risk class, which training, and who signs. That check takes thirty minutes. It prevents the policy from lagging behind the work.
Frequently asked questions
Must we take a specific course to comply with article 4?
No. The regulation prescribes no fixed course. It concerns measures and being able to show them. For most teams a training per role plus a register is a sound starting point.
Are ChatGPT or Copilot forbidden if we follow the GDPR?
No. It concerns the account, the data and the agreements with the vendor. With a business account, a data processing agreement and a policy that states which data may not go in, a team can use the tool within the rules.
Does this module replace legal advice?
No. The module gives the framework and the working method. For a specific use, a DPIA or a contract, consult a lawyer or your data protection officer.
Module 1
AI basics for teams
What a language model does, where it goes wrong, and what may go into it. For the whole team, not only the technologists.
Module 2
Working with ChatGPT, Copilot and Claude
Prompts that work, your own assistants built on your own documents, and safe use in the business version.
Module 3
Finding processes that are worth AI
Process mapping, working out the return and choosing the one process that becomes a system first.
Module 4
Automating with n8n, Make and Power Automate
Triggers, actions and conditions. Which platform fits which work, and how an automation survives the week.
Module 6
From pilot to production
Why most pilots do not land, and what a team must arrange before something runs daily.
Module 7
The AI-native organisation: leadership and roles
What changes in roles, decisions and the operating model when AI carries daily work.
Module 8
Building agents for operations
An agent that completes a multi-step task, with limits, checks and a person at the end.
Which team needs this module?
For half day €2,450 excl. VAT. In the conversation we say whether the module fits, or whether a build day or the adoption programme is better.