← Back to Scriba
Scriba · Legal

AI Usage Policy

Last updated: July 2026 · Scriba by Consilium Systems · Sydney, New South Wales, Australia

This AI Usage Policy explains how Scriba, built by Consilium Systems (ABN: 94 655 810 737), uses artificial intelligence, what that means for the outputs you receive, and what is expected of you as a licensed adviser using the platform.

This policy supplements, and should be read alongside, our Terms of Service and Privacy Policy. Where this policy and those documents overlap, the Terms of Service govern any conflict.

01

How Scriba Uses AI

Scriba uses Anthropic's Claude API to assist with drafting Statements of Advice (SOAs) and running compliance checks against the content you provide. When you use the platform, the information you enter — fact-find data, meeting notes, client circumstances — is sent to Claude, which generates draft text, structured compliance findings, or extracted data based on that input.

Scriba does not use a general-purpose chatbot interface. Prompts and instructions sent to the AI are fixed, server-side, and scoped to specific drafting and compliance tasks — you are not able to send arbitrary instructions to the underlying model.

02

What AI-Generated Outputs Are — and Aren't

Every output Scriba produces — SOA sections, compliance findings, extracted client data — is a draft starting point, generated by a language model. It is not:

Mandatory human review: Every output generated by Scriba must be reviewed, verified, and approved by the responsible licensed adviser before it is used with, or provided to, a client. You remain solely responsible for the accuracy and compliance of any document you issue under your AFSL.

03

Known Limitations

Like all AI language models, the underlying model can occasionally produce plausible-sounding but incorrect statements ("hallucinations"), miss context you assumed was implied, or apply general knowledge in ways that don't match current legislation or product-specific detail. Treat AI-generated compliance findings and figures as prompts for your own checking, not as a final compliance sign-off.

04

What Data Is Used, and How

Inputs you provide are sent to Anthropic's Claude API for processing and are not used by Consilium Systems or Anthropic to train AI models without your explicit consent. See our Privacy Policy for full detail on data storage, retention, and overseas disclosure.

You should not enter sensitive personal information beyond what is reasonably necessary to generate the specific output you're producing.

05

Your Responsibilities

06

Regulatory Context

Scriba is designed to operate within existing Australian regulatory frameworks, including the Corporations Act 2001 (Cth), the Australian Privacy Act 1988 (Cth), and your obligations as an AFSL holder or authorised representative. As Australian AI-specific standards and legislation develop — including work currently underway through the Australian Government's Office of AI — this policy will be reviewed and updated to reflect any new obligations that apply to Scriba or to your use of it.

07

Changes to This Policy

We may update this AI Usage Policy as the platform, the underlying AI models, or the regulatory environment evolves. The most recent version will always be available at scriba.consiliumsystems.com.au/ai-usage-policy. Continued use of Scriba following any update constitutes your acceptance of the revised policy.

Questions about how we use AI?

If you have any questions about this AI Usage Policy, please contact us directly. We aim to respond within one business day.

Consilium Systems
Sydney, New South Wales, Australia
info@consiliumsystems.com.au

Contact Us →