Grade AI Use Policy

At Grade, we use AI to improve the quality and speed of our work. Our position is simple:
AI guided by human judgement, not hype.

Whatever helps produce a piece of work, a person remains accountable for it.

We've set out how we work with AI so our use stays safe, honest and deliberate as we adopt more tools — so we gain the benefit while managing the real risks.

Our principles

• Clear benefit. We use AI where it genuinely helps, not for its own sake.

• Human accountability. For anything consequential, a person owns the decision and the output.

• Privacy, IP and security by design. We protect our data, our clients' data, and third parties' rights.

• Honesty with clients. We're transparent about how work was produced where it's material.

• Proportionate controls. More care for higher-stakes uses, a lighter touch for low-risk ones.

• Environmental responsibility. AI carries a real energy cost, so we use the right-sized tool for the task, avoid needless repeated runs, and prefer providers with credible climate commitments where there's a genuine choice — consistent with our B Corp and carbon-neutral commitments.

The hard lines

These are non-negotiable for everyone at Grade:

1. No confidential, client, or IP-protected data goes into public or unapproved AI tools.

2. All AI-assisted work is reviewed before it's shared — nothing goes out unchecked, and raw AI output is never passed off as considered work. Client-facing work also requires sign-off.

3. No AI makes a consequential decision without an accountable human in the loop — anything with real consequences for a person, a client, or the business.

4. We respect third-party IP and licences — confirming copyright and licence status before publishing AI-generated content.

5. Nothing illegal. No use that breaches Australian law.

How we keep it controlled

We work only from an internally approved set of AI tools, each reviewed for data governance, licensing, and where our data goes. New tools are off by default until they've been reviewed and approved. Everyone is responsible for verifying AI output, keeping sensitive data out of unapproved tools, and treating AI as an assistant rather than an authority.

Being open with clients

Where AI involvement is relevant or material to a client's work, we disclose it. When in doubt, we choose transparency.

Standards

Our approach is informed by the Australian Government's Voluntary AI Safety Standard, ISO/IEC 42001, the NIST AI Risk Management Framework, and the Privacy Act 1988.

This reflects Grade's current approach to AI. We review it at least once a year, and sooner if something material changes.