Version 1.0 · 5 November 2026 · Draft for legal and technical review
AQI AI Usage Policy
This policy describes intended AI safeguards. It is not a guarantee that every AI integration currently implements every control. Actual model providers, retention settings and technical enforcement must be verified before launch.
1. Purpose
AQI AI is an assistive feature intended to help users understand information, search records they are authorised to access, summarise material, prepare drafts and navigate permitted workflows.
2. Permission boundaries
- AQI AI must operate within the signed-in user's existing permissions, active Organisation, branch and enabled modules.
- It must not retrieve or reveal another Organisation's data.
- It must not grant access, change roles, bypass approvals or perform an action the user is not authorised to perform.
- AI access and any resulting action should be auditable where the relevant feature supports it.
3. Human responsibility
AI output can be wrong, incomplete, biased or outdated. Users must verify important information against current authoritative sources, records and applicable SOPs. A qualified human professional remains responsible for clinical and regulated decisions.
4. Restricted actions
AQI AI is not authorised to independently diagnose, prescribe, approve or release batches, sign GxP records, authorise controlled-drug transactions, make final dispensing decisions, or finalise other regulated approvals. It may prepare a draft or highlight a concern for an authorised human to review where that feature is enabled.
5. Sensitive information
Users must not submit personal, patient, confidential or commercially sensitive information unless they are authorised and the feature is approved for that data category. Organisations must determine lawful basis, provide notices and apply appropriate safeguards before using AI with personal or health information.
6. Transparency and correction
AI-generated content should be identifiable as AI-assisted where relevant. Users should be able to review, correct or reject drafts before they are used in operational records. Where AI cannot identify a reliable source, it should state the limitation rather than invent one.
7. Data handling
AQIENTRA must verify its AI vendors, model configurations, data retention and training-use settings, contractual terms and transfer safeguards. The Platform must not represent that tenant data is excluded from model training or retained for a particular period until those controls have been confirmed for the selected provider.
8. Reporting issues
Report unsafe, inaccurate, discriminatory or privacy-invasive AI output through the Platform's designated support channel. Do not rely on AQI AI for emergencies or time-critical medical decisions.