Recording a Call, With Consent You Can Take Back

Not a banner that tells you it's happening — a real yes that has to come first, and that you can withdraw later

If you've ever been on a call where a little banner said "this meeting is being recorded," you already know the polite version of consent: a notice. Someone tells you it's happening. Whether you actually agreed — and whether you could change your mind — is a different question, and usually the answer is "not really."

This is a walk through a different model, the one that shipped for NAOMS calls on 2026-04-26: an end-to-end one-to-one recording lifecycle with consent and revoke. The idea in one line: recording can't start without the other person's consent, and that consent can be revoked. Let's unpack what each of those words actually means, with no prior knowledge assumed.

A note on freshness (we promise to be honest about this): the calling work is still in progress as of mid-2026 — real one-to-one calls and this recording lifecycle shipped, but the feature isn't fully closed, and the recording surface kept evolving after April (a recordings viewer, decrypt-and-playback verification, and more landed in May). So treat this as how call recording with consent worked as of late April 2026 — the principle is stable; specific buttons and screens may have grown since. We'd rather tell you that than quietly let a tutorial go stale.

The three words that matter

Lifecycle. A recording isn't a single instant — it has a beginning, a middle, and an end, and each stage has rules. "Lifecycle" just means we treat the whole arc as one governed thing, not a fire-and-forget button. What shipped is an end-to-end recording lifecycle, which is the important part: the consent and the recording and the ability to stop are all one connected flow, tested from start to finish against a real call.

Consent. This is the load-bearing word. In this model, consent is not a notice you're shown after the fact — it's a condition that must be true before recording can begin. Think of it like a door that only opens when both people turn their key. If the other person hasn't agreed, the recording simply does not start. There's no "record now, ask forgiveness later." The agreement is upstream of the recording, not a footnote to it.

Revoke. This is the word that makes the consent honest. A consent you can never take back isn't really consent — it's a one-time signature on an open-ended permission. Here, consent is revocable: the person who agreed can later withdraw that agreement. Revocation is a first-class action, shipped in the same lifecycle, not an afterthought you have to email someone about.

Why build it this way?

Because the alternative quietly assumes the recorder's convenience matters more than the other person's agency, and a local-first system is built on the opposite assumption. The whole premise of NAOMS is that you control what happens to information about you. A recording of your voice is information about you. So the system that produces it has to put your agreement at the front (you can refuse) and your ability to change your mind at the back (you can revoke). Anything less would contradict the thing the whole project exists to do.

There's also a plain practical honesty to it. A notice-only model puts the burden on you to object in the moment, fast, often awkwardly. A consent-required model puts the burden where it belongs: on the person who wants to record, to actually get a yes first.

What to take away

You don't need to remember any commands to understand the point of this one, so we'll leave you with the shape instead of a screen:

Recording is a lifecycle, governed start to finish — not a single button press.
It begins only with consent — a real yes from the other person, checked before recording starts, not a banner shown after.
That consent is revocable — the agreement can be withdrawn later, as a real, supported action.

When you record someone in a system built this way, you're not informing them. You're asking them — and leaving the door open for them to change the answer. That's the difference, and it's the whole point.

Written by AI agents from real project logs; owned and edited by Mujo.