AI Transparency

Gluten Zero uses AI to help explain food decisions — not to replace clinical judgement. Here is exactly how it works and where its limits are.

Evidence-first architecture

Before any AI model generates a response, our Assistant Response Router queries structured knowledge — ingredient libraries, food-risk intelligence, and FSANZ regulatory rules. AI supplements evidence; it does not override it.

Safety guardrails

• Forbidden certainty phrases (e.g. "100% safe", "guaranteed gluten-free") are blocked.
• Medical diagnosis and treatment requests are declined with appropriate referral.
• Low-confidence answers trigger uncertainty messaging, not guesses.
• Provider failover chain with logging for audit.

Data handling

When you use the AI assistant, your question and recent chat turns may be processed by third-party AI providers (Google, OpenAI, Anthropic, Mistral, and/or OpenRouter) that may operate outside Australia. OCR text extraction runs on-device first; if you use cloud analysis, ingredient text — and optionally a label image you choose to upload — may be sent to our servers and to vision-capable providers. See Privacy Policy.

Human review

Published knowledge hub articles undergo human and clinical review. Real-time AI chat responses are guardrailed but not individually reviewed before display — always verify critical decisions with packaging and your care team.

Responsible AI commitment

We do not mass-generate low-quality SEO content using AI. Every published article must pass our Editorial Policy and AI content quality checklist.