AI assistant

Reviewing drafts, data terms and privacy

Everything the assistant writes is a draft to review, what is sent for processing, and the data terms the owner accepts

Last updated:

Drafts need your review

Everything the assistant produces is a draft that may contain errors, gaps or inexact citations. A lawyer must review it and verify citations and regulations before any use; professional responsibility stays with the lawyer. So:

  • You see "Everything the assistant writes is a draft for review; check it before relying on it."
  • A saved document carries "Draft by the AI assistant — review before use".
  • The assistant saves nothing to your matter on its own; saving happens when you press, after review.

What is sent

When you run a feature, only the data needed for that action (the text of the matter, document or notes you choose) is sent to Anthropic, the AI model provider. The matter data sent excludes phone numbers, ID numbers and emails. Legal research does not send your clients' data.

The data-processing terms

The firm owner accepts them once under Settings then AI assistant with I accept the terms, after a confirmation. They include:

  1. Only the data needed for the action is sent to Anthropic.
  2. Anthropic does not use the data to train its models, and neither do we.
  3. Processing happens outside the Kingdom and counts as a cross-border transfer under the Personal Data Protection Law. The firm is responsible for informing its clients where required.
  4. Outputs are drafts to be reviewed.
  5. Every action (who, when, points) is logged where the owner can see it.

The assistant does not work for any user until the owner agrees. Others see "The firm owner must accept the AI data-processing terms first."

Was this article helpful?

Still need help?

Write to us and support will usually reply by email within one working day.

Contact support