How to Transcribe Research Meetings with AI under GDPR (2026)

The short answer

Tell everyone before the meeting that it will be recorded and transcribed, by which kind of service, and why. Give people a real chance to object. Use a service that processes data in the EU and does not train on it, keep transcripts only as long as they are useful, and share summaries rather than full transcripts where you can.

Two kinds of research meetings

Working meetings (lab meetings, supervision, project meetings) are about running the research. The transcript is an internal record. At a public university the usual GDPR basis is public task (Article 6(1)(e)), while private institutions may use legitimate interest. Either way, people need to be informed and colleagues should be able to say no.

Data-collecting meetings (stakeholder workshops, expert panels, Delphi rounds held live, community meetings) produce research data. Treat them like interviews or focus groups: information sheets, consent and ethics approval where required.

Good practice for recording meetings

Announce recording in the invitation and again when the meeting starts.
Do not record supervision or one-to-one meetings without the other person’s clear agreement. Supervision often touches health, stress or personal matters.
Pause recording for personal or confidential items.
Check that guests from other organisations agree, since their own employer may have rules.
Avoid meeting bots that join automatically and send audio to services you have not checked.
Keep the transcript in the project it belongs to, not in personal inboxes.

Making transcripts useful

A raw transcript of a one-hour meeting is around 9,000 words. Nobody reads that. What helps is a short summary with decisions, open questions and actions, linked back to the moment in the recording where each was discussed.

In Kahubi you can record a meeting in the browser or upload a recording, get a transcript with speaker labels processed in the EU, and ask the agent for decisions, action points or a summary for colleagues who missed it. Because the transcript lives in your research project, you can also ask how today’s discussion relates to your notes, drafts and papers.

A note for PhD students and supervisors

Recording supervision can help students remember feedback and reduce misunderstandings. Agree on it together, decide who keeps the recording, and delete it once the notes are written. A shared summary of decisions is usually more useful than a full transcript, and less sensitive.

Sources

Last updated 2026-10-09.

Frequently asked questions

Do I need consent to record a research meeting?

Everyone must be informed before recording starts and should have a real chance to object. For working meetings the GDPR basis is often public task (or legitimate interest at private institutions) rather than consent. For meetings that collect research data, such as workshops with stakeholders, follow your normal consent and ethics procedures.

Are meeting bots GDPR compliant?

It depends on the provider, where it processes audio and whether it trains on it. Bots that join calls automatically often process data outside the EU. Check them with your data protection officer before use.

How long should I keep meeting transcripts?

Only as long as they are useful. For working meetings, keep the summary and decisions and delete the full transcript and recording after a set period. For data-collecting meetings, follow your data management plan.

How to transcribe research meetings with AI under GDPR