AI Transparency Statement
Last updated: 16 March 2026
This statement explains how Internest uses artificial intelligence, consistent with the OAIC's October 2024 guidance on AI and privacy and Australia's Voluntary AI Safety Standard.
How we use AI
Bill processing
When you upload a bill, AI (Anthropic's Claude, via their commercial API) extracts numerical data — plan names, rates, usage figures, and dates — through OCR and natural language processing. Original files are processed in real time and deleted after extraction.
Comparison calculations
Our comparison engines are deterministic calculation engines, not machine learning models. They use fixed formulas to compute annual costs, savings, and projections. AI assists in structuring and presenting results, but the core maths is rule-based and auditable.
Content generation
Some educational and informational content on the platform is AI-assisted. This includes page descriptions, methodology explanations, and social media posts. All AI-generated content is reviewed for accuracy and regulatory compliance before publication.
Operational automation
Our Ember automation system handles routine operational tasks: data freshness monitoring, backup verification, and social media scheduling. These systems use AI for execution but operate under strict human-defined rules and guardrails.
Future: AI assistant (Phase B3)
A planned conversational AI assistant will help users understand their bills, ask questions about comparison results, and navigate the platform. This feature is not yet live.
What AI does not do
- Does not provide personalised recommendations. AI does not consider your objectives, financial situation, or needs when generating results. All outputs are general information.
- Does not replace professional advice. AI outputs are not financial, tax, health, legal, or insurance advice. Always consult a qualified professional for important decisions.
- Does not use your data for model training. Anthropic's commercial API terms explicitly prohibit using customer data for model training. Your queries, bill data, and medication lists are not used to improve AI models.
- Does not make autonomous decisions. AI assists with data processing and calculation — it does not independently decide what products to show you or how to rank them.
Where your data goes when using AI features
When you use AI-powered features, relevant data is sent to Anthropic's API servers in the United States for processing. This constitutes a cross-border disclosure under APP 8 of the Privacy Act.
| Protection | Detail |
|---|---|
| No model training | API customer data is not used for training or improving Anthropic's models |
| Limited retention | Anthropic retains API data for a maximum of 7 days, then deletes it |
| Encryption in transit | All data is encrypted via TLS |
| Trust & safety | Anthropic may flag content that violates their usage policies for safety |
Data sources by vertical
Transparency about where our comparison data comes from:
| Vertical | Primary data source |
|---|---|
| Energy | Australian Energy Regulator (AER) CDR Product Reference Data APIs |
| Gas | AER CDR Product Reference Data APIs (fuelType=GAS) |
| Insurance (Health) | privatehealth.gov.au (PHOL Schema V3.1) |
| Insurance (Pet/Car/Home/Life) | Provider websites, APRA/ASIC published data |
| Banking | CDR Product Reference Data APIs (121+ ADIs), APRA published data |
| Medicine | PBS Schedule data (Australian Government), TGA public databases |
| Super | APRA MySuper Product Heatmap data, quarterly reports |
| Telecom | Provider websites, ACCC Measuring Broadband Australia public data |
| Subscription | Provider websites, CDR merchant mapping |
AI limitations
AI-generated results may be inaccurate, incomplete, or outdated. Specifically:
- Savings estimates are approximations based on available data and your inputs
- AI may misread or misinterpret uploaded documents
- AI does not have access to real-time pricing and may reflect outdated information
- AI outputs should be treated as a starting point, not a definitive answer
Always verify important information directly with the relevant product provider before making decisions.
Human oversight
Internest operates under a “1 Human + N AI” model — a solo founder working with AI systems as execution partners. This means every AI-generated output, every data pipeline, and every calculation engine has a human accountable for its correctness and compliance.
AI features are designed with guardrails to prevent the generation of advice or recommendations. We regularly review AI outputs for accuracy and regulatory compliance. If you believe an AI output is incorrect, please contact us.
Your choices
You can use many of our comparison tools without AI processing by entering information manually rather than uploading bills. Our core comparison engines are deterministic calculators that do not require AI to function.
Contact us
Questions about our use of AI?
Email: info@internest.com.au
Website: internest.com.au