Trust & control

AI permissions and the safe sandbox

KolWrite's AI assistant can do real work for you — transcribe and caption recordings, clean up and edit transcripts, organize your media, and handle everyday steps. Permissions decide when it acts on its own and when it stops to ask you first. This page explains, in plain language, how those controls work and how your device keeps the assistant safely contained.

The short version

  • KolWrite runs as an ordinary app on your device. Your device's operating system is the sandbox — the boundary that decides what any app, including KolWrite, is allowed to touch.
  • Permissions inside KolWrite control when the assistant asks you before acting. They never expand what KolWrite is technically able to reach on your device.
  • Some actions always ask for your approval — no setting turns those confirmations off.
  • You can change or withdraw permission at any time, on each device, in Settings.

The safe sandbox is your device, not KolWrite

When KolWrite's assistant does something, that action happens inside the access your operating system has already granted the KolWrite app — nothing more. macOS, Windows, Linux, iPhone, iPad, Android, and your web browser each enforce their own protections and permission prompts, and KolWrite works within them.

Two things matter here:

How permissions work

KolWrite gives you a simple choice for how much the assistant does on its own.

Ask me first (recommended, default)

The assistant can transcribe, search, and prepare edits freely, but it checks with you before anything that changes your transcripts, your files, or anything else sensitive. You see exactly what it is about to do, and you approve it.

Full access

The assistant carries out routine actions without asking each time, so it can keep moving on longer jobs. Choose this only for work you trust it to handle. Even with full access, the actions in the next section still ask first.

Where KolWrite offers an always-on assistant, you can also tune how much initiative it takes — from suggestions only, through asking before every change, up to handling routine steps on its own. Any more automatic setting is an explicit, opt-in choice, and mandatory safety confirmations always remain in force.

What always asks first

No permission level removes these confirmations. Whatever mode you choose, KolWrite still stops and asks for a one-time approval before anything that is:

You always see the exact change before it happens.

Why it works this way

Asking before every small step gets tiring, and constant prompts make it hard to notice the ones that matter. Letting the assistant handle routine work inside a clear boundary means fewer interruptions — and a trust model you can actually reason about. You are not only relying on the assistant's judgment; you are relying on limits your device enforces, and on confirmations that cannot be switched off for the actions that carry real consequences.

Across your devices

KolWrite works the same way on every platform, and the outer boundary is always your device's own operating system:

Permissions are per device. Choosing a more automatic mode on one device does not change anything on your other devices, and revoking a system permission on a device stops it there.

You stay in control

You can review, change, or withdraw the assistant's permissions at any time in KolWrite's settings, and you can adjust the underlying device permissions in your operating system's own settings. More automatic modes are opt-in and always reversible.

You remain responsible for actions you ask the assistant to take, or allow it to take on your behalf — including under Full access. AI can make mistakes, so review anything important before you rely on it. This page explains how the controls work in plain language; it is not a contract. Your use of KolWrite is governed by our Terms of Service and Privacy Policy.