Ethics, privacy & responsible AI
KallosSim, VitalsVoice, and Reheara use the Praxis AI engine. Our purpose is to support conversation practice and reflection, with people retaining professional judgement.
This section brings together the principles and governance topics documented for KallosSim. It is an overview, not a replacement for a product privacy notice, organisational agreement, or legal assessment. Sharing an engine does not mean every application shares the same data flows, retention settings, or compliance status.
Ethics, fairness & wellbeing
Practice should help people prepare for difficult conversations without putting real service users at risk.
- Use fictional scenarios for rehearsal; keep real client identities and live case records out of practice.
- Challenge stereotyping and discrimination. Protected characteristics must not become shortcuts for explaining a person's situation.
- Recognise that difficult fictional content can still be distressing. Learners should be able to stop, skip, report concerns, and seek supervision or debrief.
- Treat guardrails and reporting as safeguards, not guarantees of perfect fairness or human pre-approval of every generated session.
Privacy & data rights
People should understand what is processed, why it is needed, who can access it, and how to exercise their rights.
- Minimise personal data and make account data, practice records, organisational access, and provider processing understandable.
- Explain access, correction, deletion, portability, restriction, and objection routes in the relevant product's privacy notice.
- Separate transcript retention, scores and session metadata, provider retention, and optional research or model-improvement consent. These are different data uses.
- Use product-specific retention, sub-processor, international-transfer, and consent information. KallosSim's policy does not automatically describe VitalsVoice or Reheara.
AI transparency & human judgement
Be clear about what AI generates, what humans design, and where the system can be wrong.
- Distinguish AI-generated scenarios, dialogue, and feedback from human-maintained learning criteria, prompts, and professional frameworks.
- Make AI involvement and simulated voices clear. KallosSim documents user disclosures and machine-readable markings for generated content.
- Treat scores as formative learning aids. They do not replace supervision, professional judgement, or formal assessment of competence.
- Acknowledge that outputs may be biased, inconsistent, unrealistic, or inappropriate. Provide a route to report harmful output or challenge feedback.
EU AI Act & organisational governance
Intended use, human oversight, AI literacy, and evidence matter more than a blanket compliance badge.
- KallosSim's documented intended use is formative practice, not examinations, grading, certification, hiring, promotion, dismissal, or performance management.
- Its transparency materials address AI literacy and Article 50 disclosures and marking. Organisations still need to brief users and review their own use and data-protection responsibilities.
- KallosSim maintains classification and assurance documentation. The classification memo reviewed for this section is unsigned; internal preparedness is not completed conformity assessment.
- KallosSim's published materials state that it is not CE marked and is not registered in the EU AI database. Shared engine mechanics do not establish identical legal classification or compliance across products.
Questions, concerns & procurement
For Cortavia, VitalsVoice, Reheara, or Praxis enquiries, contact hello@cortavia.com. For KallosSim privacy requests, use dpo@kallossim.com. KallosSim harm reports can be sent to hello@kallossim.com with the subject “AI Act incident”.
Before organisational adoption, review the relevant product notice, intended use, retention schedule, processor arrangements, access controls, incident route, and any DPIA or local approval required for your deployment.