ProductUse casesPricingBlogContact
Dashboard Sign in Start free
Artificial Intelligence in Business

Chatbot GDPR Checklist: What to Settle Before Launch

Chatbot GDPR Checklist: What to Settle Before Launch

Before you launch a website chatbot, settle a short list of questions: what personal data the chat collects, on what lawful basis, who processes it, where it goes, how long it stays, how people can exercise their rights and how you tell them they are talking to an AI. And chatbot GDPR work falls on you, because the business that puts the bot on its own site is the controller and carries the responsibility, not the visitor typing into the widget. The checklist below follows launch order. It starts with mapping the data and ends with the AI disclosure.

What personal data can a chatbot conversation contain?

Pretty much anything. A chat can hold any personal data a visitor decides to type, plus technical data such as an IP address or session identifiers. The usual suspects: names, email addresses, phone numbers, order or booking details, account questions, and whatever people volunteer in free text (and they volunteer a lot). For each type, write down why you process it. Answering questions? Following up? Improving answers? Then pick and document the lawful basis under GDPR Art. 6 that fits each purpose, before you touch a single setting.

Who is the controller, and why you need a chatbot DPA

In most setups the website owner is the controller. The chatbot vendor and the AI model provider act as processors or sub-processors. If you’re unsure where the lines fall, the European Data Protection Board walks through it in its guidelines on controller and processor. Now, the vendor handles AI chatbot personal data on your behalf, so GDPR Art. 28 requires a data processing agreement. No DPA, no launch. Check that it lists every sub-processor, the model provider included, then keep a signed copy and note the signing date in your records.

Where is chat data stored, and does it leave the EEA?

Ask the vendor where conversations are stored and processed (the AI model provider counts here too) and whether any of that happens outside the EEA. If data goes abroad, you need a transfer mechanism under GDPR Chapter V (Art. 44 onwards), such as an adequacy decision or standard contractual clauses. Questions to put to a chatbot vendor about GDPR before signing anything:

  • Where are conversation logs stored, and in which country are they processed?
  • Which sub-processors receive chat data, including the AI model provider?
  • What transfer mechanism covers any data leaving the EEA?
  • Can you set or shorten the retention period yourself?
  • How do you delete one person’s data on request?
  • Are conversations used to train models, and can that be switched off?
  • Is data encrypted in transit between the widget and the servers?

Chatbot data retention: how long logs stay and who deletes them

Pick a fixed retention period for conversation logs that matches the purpose. That is the storage limitation principle in GDPR Art. 5. Then decide who actually deletes. Does an automatic setting handle it, or is it a named person’s job? And how often does someone check that it really happened? Backups and exported transcripts count too. People forget those all the time, so put them under the same rule. Write the period down, because your privacy notice will have to state it anyway.

Limit what users type and what the bot knows

The safest personal data is the data that never enters the chat. So tell users, plainly, not to share sensitive information. A short line inside the widget does the trick: please don’t type health data, ID numbers or passwords here. Same logic on the other side of the bot. Keep sensitive internal documents out of its knowledge, and read our guide on what never belongs in sources before you upload anything. Review those sources before launch. Then again after every update.

Data subject rights and information duties

When someone asks, you have to be able to find, hand over and delete that one person’s conversations, under the rights of access and erasure in GDPR Art. 15 and 17. Sounds simple. Is it, in your tool? Test the process before launch: search logs by email or session, export the result, delete it, confirm the deletion. Article 13 brings information duties on top, so your privacy notice needs the chat’s purpose, lawful basis, recipients, transfers and retention period. Our guide to chatbot privacy notice wording goes through it line by line.

Pre-launch recap for a GDPR compliant chatbot:

  1. Data map with a purpose for each data type
  2. Documented lawful basis
  3. Signed DPA listing sub-processors
  4. Transfer mechanism for any data leaving the EEA
  5. Retention period and a named owner for deletion
  6. Input limits in the widget and clean knowledge sources
  7. Tested procedure for access and erasure requests
  8. Updated privacy notice
  9. AI disclosure in the chat window

Tell users they are talking to an AI

Article 50 of the AI Act says people must be informed when they interact with an AI system, unless that is obvious from the context. So say it, clearly, right in the chat window. The full text is in Regulation (EU) 2024/1689 on EUR-Lex, and our article on AI Act chatbot disclosure covers the practical wording.

Go through the chatbot GDPR questions in this order and you’ll have documented answers before the first visitor opens the widget. Which is exactly what a controller is expected to have. One honest caveat: this is a checklist, not legal advice. Wherever your data, your sector or your transfers raise questions it can’t settle, bring in your data protection officer or a lawyer.

For a Botino assistant, several of those answers are already documented: the bot draws only on files you upload, conversations go over encrypted HTTPS, and you choose what enters the knowledge base.

FAQ

Is a chatbot GDPR compliant out of the box?

No. No tool makes a website compliant on its own. Compliance comes down to how you, as the controller, configure the bot, sign the right agreements and document your choices. A well-built product makes that easier, sure. But responsibility for chatbot data protection stays with the site owner.

Do I need a DPA with my chatbot provider?

Yes, whenever the provider processes personal data on your behalf, and GDPR Art. 28 requires that relationship to be governed by a contract. The agreement should also cover sub-processors such as the AI model provider. Keep the signed copy with your processing records.

How long should I keep chatbot conversations?

GDPR doesn’t set a fixed period. Under Art. 5, you keep logs only as long as the purpose requires, then delete them, backups included. Document the period you choose and state it in your privacy notice.