# AI & data privacy

> How your data is (and isn’t) used by Kepler’s AI features.

_Collection: AI & Automation (ai-automation) — Kepler Help Center. Canonical: https://keplercrm.com/support/articles/ai-data-privacy/_

## The short version

  - **Kepler processes workspace data to provide the AI feature you asked for** — parsing structures *your* database; matching ranks *your* candidates against *your* jobs; the Assistant answers from *your* records.

  - **Access controls apply** — AI never becomes a side-channel around your sharing settings.

  - For current subprocessors, retention, training-use and compliance details, contact your workspace administrator or Kepler support.

## How processing works

When an AI feature runs, the relevant content (a CV, an email, a transcript) is processed to produce its output — structured fields, a match verdict, minutes, a proposal — and the output is stored in your workspace like any other data you created. Processing is workspace-scoped: nothing from your workspace informs results in anyone else's.

## Visibility is respected

AI features inherit your access rules. Insights derived from restricted mail (see [Email privacy & sharing](/support/articles/email-privacy-and-sharing/)) are only shown to people allowed to see that mail; AI never becomes a side-channel around sharing settings. The Assistant answers only from records you could open yourself.

## Control and audit

  - AI-suggested changes are shown in [Notifications](/support/articles/ai-suggestions/) before they are applied.

  - Suggested updates can target centrally supported built-in fields and custom columns configured to read from source content.

  - Users with permission to delete activities can permanently remove a meeting's recording, transcript and minutes using **Delete meeting data**.

  - Candidate data handling — access, export and erasure — is covered in [Security & compliance](/support/articles/security-and-compliance/).