Personal data handled properly, also when you use AI

Almost every website, web shop and business application processes personal data: names, e-mail addresses, orders, sometimes medical or financial data. That is covered by the GDPR, called the AVG in the Netherlands. With the rise of AI comes a new question: what happens to personal data when you put it into ChatGPT, Claude, Gemini or an AI application of your own? ClickSmash helps you get that right technically, in the software we build and in the systems you already have.

In short

  • The GDPR requires you to process no more data than needed, and to secure it well.
  • Whatever you paste into an AI tool leaves your organisation: handle it with a business account, a data processing agreement and less personal data.
  • Since August 2026 the AI Act requires transparency about chatbots; high-risk requirements follow from December 2027.
  • We take care of the technology; you do the legal assessment with your lawyer or DPO.

The GDPR in short

The GDPR applies to every organisation that processes personal data. The main principles:

  • Legal basis and purpose: you only process data for a valid reason, such as a contract or consent, and only for the purpose for which you received it.
  • Data minimisation: you collect no more than needed, and keep it no longer than needed.
  • Security: you take appropriate technical and organisational measures, such as encryption, access control and logging.
  • Rights of data subjects: people may see their data, have it corrected or have it deleted.
  • Accountability: you must be able to show you have it in order, with a record of processing, data processing agreements and a DPIA where needed.
  • Data breaches: a data breach that poses a risk to the people concerned is reported within 72 hours to the data protection authority, in the Netherlands the Autoriteit Persoonsgegevens.

Data minimisation in practice

The safest data is data you do not have. In software that means:

  • Forms only ask for what is needed: no date of birth if you do not use it.
  • Data is deleted or anonymised automatically after the retention period.
  • Test and development environments work with fictitious or pseudonymised data, not with a copy of production.
  • Staff only see the data they need for their work.
  • Log files, analytics and backups contain no more personal data than needed, and have a retention period too.

Data processing agreements

If another party processes personal data for you, such as a hosting company, an e-mail service, an accounting package or an AI supplier, that party is a processor and you sign a data processing agreement. It states which data is processed, for what, how it is secured, which sub-processors there are, where the data is kept and what happens in case of a data breach. We help you map which suppliers and integrations can reach your data, and where an agreement is missing. For the hosting and management we do ourselves, we sign one with you.

DPIA: mapping privacy risks in advance

A DPIA (Data Protection Impact Assessment) is required when processing is likely to pose a high privacy risk, for example health data on a large scale, tracking or profiling people, or new technology such as AI that makes decisions about people. You describe what you process and why, assess the risks for the people concerned and record which measures reduce them. We deliver the technical part: which data is kept where, who can reach it, how it is secured and which measures are possible in the software. You do the legal assessment with your lawyer or data protection officer; we are not lawyers.

AI and privacy

AI tools are useful, but whatever you paste into a prompt leaves your organisation. A few questions come up:

  • Where does the data go? With a free or personal account, what you enter may be used to train the model, and the data is often kept outside the EU. Business plans and APIs from Anthropic, OpenAI and Google among others offer agreements that your data is not used for training, with a data processing agreement.
  • Which data goes in? Personal data, customer files and confidential documents do not belong in an AI tool without thought. Where possible, remove or pseudonymise them before they reach a model.
  • Who has access? An AI assistant that can search your documents or systems may only show what the user is allowed to see themselves.
  • What does the AI say back? Language models can give incorrect information about people. Decisions about people need a human who checks.

So set agreements in an AI policy: which tools staff may use, with which account, and which data may and may not go in. See also what the Dutch Data Protection Authority writes about algorithms and AI (in Dutch).

The EU AI Act

The EU AI Act regulates AI by risk, alongside the GDPR. It has been in force since August 2024 and applies step by step:

  • Prohibited practices, such as social scoring and emotion recognition in the workplace, have been banned since 2 February 2025.
  • Transparency: since 2 August 2026 you have to tell people they are talking to an AI chatbot, and AI-generated content must in certain cases be recognisable as such.
  • High-risk AI, such as AI for recruitment, credit scoring or access to education, faces strict requirements. Following the EU Digital Omnibus, those apply to most uses from 2 December 2027.

For most SMEs it is mainly about transparency and the GDPR: which personal data your AI application uses, on what legal basis, and how you protect it.

How ClickSmash helps

  • Mapping: which personal data your systems process, where it is kept, who can reach it and which suppliers and integrations work with it. That is the basis for your record of processing and data processing agreements.
  • Privacy by design: in software we build, data minimisation is part of the design from the start, with retention periods, roles and rights, encryption, logging of who views which data and functions to view, export or delete data.
  • Improving existing systems: we adapt existing applications, also when someone else built them, for example to clean up old data or test environments.
  • Using AI safely: we choose the right model for each use, remove or pseudonymise personal data before it reaches an AI model in the cloud, and can process sensitive data with a model we host ourselves in the Netherlands. Read more at AI in your business.
  • DPIA and security: we deliver the technical part of a DPIA, and test the security with an OWASP penetration test or a cybersecurity scan.
  • Data breaches: an emergency plan for what you do in case of a data breach, and logging that shows afterwards which data was accessed.

Want to talk through what the GDPR and the AI Act mean for your plans first? See software and AI advice.

Sources

Is your privacy in order technically?

We map which personal data your systems and AI applications process, and what can be improved. You get an overview with what to tackle first.

Get in touch

Or call 020 717 3360

Frequently asked questions

What is the difference between the GDPR and the AVG?

None: AVG (Algemene verordening gegevensbescherming) is the Dutch name of the European General Data Protection Regulation (GDPR).

May I put personal data into ChatGPT or another AI tool?

Not just like that. You need a legal basis, the supplier is a processor with whom you sign a data processing agreement, and you must know where the data goes and whether it is used for training. Use a business account or API, and remove or pseudonymise personal data where you can.

When do I need a DPIA?

When processing is likely to pose a high privacy risk, for example health data on a large scale, tracking or profiling people, or AI that makes decisions about people. Data protection authorities, including the Dutch one, publish lists of processing for which a DPIA is required.

What is a data processing agreement?

An agreement with a party that processes personal data for you, such as a hosting company or AI supplier. It states which data is processed, for what, how it is secured, which sub-processors there are and what happens in case of a data breach.

Do you sign a data processing agreement?

Yes. For the hosting and management we do, we sign a data processing agreement with you.

Does the AI Act apply to my business?

If you use or offer AI, yes, but for most SMEs it is mainly about transparency: people must know they are talking to an AI chatbot. Strict requirements apply to high-risk AI, such as AI for recruitment or credit scoring, and for most uses they apply from December 2027.

Can you use AI without data leaving the Netherlands?

Yes, with an AI model we host ourselves in the Netherlands. For less sensitive uses we also use business APIs from Anthropic, OpenAI and Google among others, removing personal data first where needed.

Are you lawyers?

No. We take care of the technology: which data is kept where, how it is secured and what needs to change in the software. You do the legal assessment, such as the legal basis and the text of a privacy notice, with your lawyer or data protection officer. We are happy to work with them.

What should I do in case of a data breach?

Limit the damage, record what happened and report a data breach that poses a risk to the people concerned within 72 hours to the data protection authority. If the risk is high, also inform the people concerned.