©Unseen Studio – Unsplash Artificial intelligence is increasingly becoming part of the day-to-day work of associations. Many organizations are now experimenting tools capable of automatically recording meetings, generating transcripts and producing minutes.
At first glance, these tools offer clear advantages. They can save time, reduce administrative workload and help capture discussions more efficiently. However, meeting minutes are more than simple administrative notes. They should reflect the official decisions taken by an association’s governing body and may later be relied upon in matters of governance, compliance or even litigation. As a result, the use of AI in minute-taking raises an important question: how can associations use these tools while preserving the reliability of their official records?
Traditional role of minutes
Traditionally, meeting minutes are not intended to capture every word spoken during a meeting. They are not meant to be verbatim transcripts. Instead, their purpose is to summarise the key points of the discussion, record what was decided and identify the actions that need to follow.
AI tools, however, automatically generate several types of records from the same meeting, including : a recording of the meeting, a full transcript generated by AI, an automated summary of the discussion, and the official minutes.
In a legal context, these additional documents, such as transcripts or recordings, may become discoverable and could potentially be used as evidence. If inconsistencies arise between them, questions may emerge about which document accurately reflects what happened during the meeting. Even informal comments captured in transcripts may later be interpreted out of context.
None of this means that AI tools should not be used. On the contrary, they can be very helpful when preparing draft minutes. But it does highlight the importance of carefully managing how these records are created, reviewed and retained. Ultimately, meeting minutes should remain a concise governance document focused on decisions, rather than a detailed transcript of the entire debate.
Recording meetings and legal exposure
Many AI minute-taking tools rely on recording meetings in order to generate transcripts and summaries. In practice, recording a meeting can be helpful when preparing minutes, as it allows the minute-taker to verify key points and ensure that decisions are accurately reflected.
However, permanently storing recordings of internal discussions may also create unnecessary risks. Indeed, recordings capture everything that is said during a meeting, including preliminary opinions, informal comments or ideas that were never adopted. Organizations must also ensure compliance with data protection rules by informing participants of any recording, obtaining their explicit consent, and clearly explaining the purpose, storage and retention of the data. Recordings should then be used only as a temporary drafting tool. Once the minutes have been prepared, reviewed and approved, the recording should be destroyed.
Limiting the retention of recordings reduces the amount of sensitive information available and may also encourage more open discussions during meetings, as participants may feel less comfortable expressing their views if they know that every remark will be permanently recorded.
These considerations are particularly relevant in settings where organizations apply principles such as the Chatham House Rule. Under this framework, participants are free to use the information shared during a discussion, but they should not reveal who made specific statements.
Errors and responsibility
Although AI tools are becoming increasingly sophisticated, they are not immune to errors. Transcription systems may misunderstand statements, misidentify speakers or misinterpret the meaning of a discussion. Even small mistakes can have consequences. For example, an inaccurate transcript might give the impression that a decision was taken when it was not, assign an action to the wrong person, or create confusion about the outcome of a vote.
It is therefore important to remember an essential principle: AI tools do not carry legal responsibility.
Responsibility for the accuracy of meeting minutes remains with the individuals who prepare and approve them. In most organizations, this means the secretary or designated minute-taker who drafts the minutes and the chair who ultimately signs or approves them.
AI can definitely support the drafting process, but the final document must always be carefully reviewed and approved by humans before it becomes part of the organization’s official records.
4. Practical guidelines for drafting AI-generated minutes
AI can be a useful tool when preparing meeting minutes, provided it is used carefully.
Do’s
- Use AI as a drafting tool, not as the final record;
- Ensure every draft is reviewed by a human before approval;
- Clearly label drafts, for example “AI-generated – subject to review.”;
- Establish clear version control to determine which document is the official record.
Don’ts
- Do not treat AI-generated output as the official minutes without verification;
- Do not keep meeting recordings longer than necessary for drafting the minutes;
- Avoid turning minutes into detailed transcripts of the entire discussion.
AI will likely become an increasingly common tool in meeting administration. Used properly, it can make the preparation of minutes faster and more efficient. However, one principle remains unchanged: meeting minutes are a governance document that must be clear, reliable and carefully reviewed. AI can assist in drafting them, but responsibility for their accuracy will always remain in human hands.