Should we leave the AI note-taker off for a client's confidential calls?

Three minutes into our design call, the client asked, ‘Does that little assistant keep everything I say?’ My colleague said it just made notes. I was about to answer yes to the next question when I realised neither of us had checked what happened to the audio.

We are a three-person studio. The note-taker has made our project handovers easier: the person presenting can actually look at the client instead of typing. But this call was about an unreleased product, and the client had asked us to keep the sketches within the project team. I asked to pause the assistant while we continued. My colleague did, then told me afterwards that I had made our new system sound careless.

I can see why that stung. I helped choose it after two meetings where our handwritten notes disagreed. Still, the Australian Cyber Security Centre advises businesses to review an AI vendor’s data handling and verify its outputs. We have not done that review for this tool. [Official guidance](https://www.cyber.gov.au/publication/artificial-intelligence-for-small-business).

Before the next call, I want an answer we can give without guessing. Is it reasonable to keep this client’s calls on human notes while we check, or does that put an unfair extra job on the colleague who already handles the follow-up?