We want you to understand when and how AI is used, in the spirit of Australia's AI Ethics Principles (2019), the Voluntary AI Safety Standard (2024) and the Guidance for AI Adoption (October 2025). We aim to align our practices with this guidance; this is a statement of intent, not a claim of certification.
When AI is used
- Many HAMANI agents are deterministic and run with no AI model at all — these are marked $0.
- Other features use an AI model (Anthropic's Claude) to generate a response; these are labelled with the model tier they may use.
How we try to use it responsibly
- Transparency: we tell you which features use AI and that outputs can be wrong.
- Human oversight & contestability: you review and approve agent actions; risky actions (sending, spending, contacting people) are gated for your confirmation.
- Privacy & security: personal identifiers are scanned/redacted before content leaves the boundary where practical; your content is not used to train third-party models.
- Accuracy & reliability: we test our deterministic features and mark honest limits; we do not present a simulated result as a real one.
- Fairness & wellbeing: we avoid inferring sensitive attributes; cultural/religious features are opt-in and self-selected, never assigned.
If an AI feature affects you and you disagree with a result, contact us — you can ask a human to review it.