AI Disclosure – Practicly

Last updated: 15.09.2026 Version: 1.0

1. What Practicly is

Practicly is an AI-powered medical training simulation platform for healthcare professionals. It lets you practice diagnostic conversations with AI-generated virtual patients in a safe, consequence-free environment.

Practicly is not a medical device, does not provide clinical advice, and must not be used for real patient care or credentialing.

2. You are interacting with AI

Every element of a virtual patient — spoken utterances, voice, avatar gestures, and written feedback — is generated by AI. There is no human on the other side of the conversation. Sessions are simulations designed exclusively for skills practice.

3. Which AI systems we use

We work with specialist sub-processors for these components under data-protection agreements. A current list of our sub-processors is available on request — see our Privacy Policy.

4. What we do with your data

• Voice is transcribed in real time; the audio stream is not stored beyond the session.
• Transcripts are retained for up to 24 months so you can review your sessions.
• Your session data is not used to train the AI providers' models. We rely on each provider's no-training API terms and enable the strongest data-retention controls our providers offer. Formal data-processing agreements are in place with each provider.
• You can request deletion of your transcripts at any time — see your Privacy Policy for GDPR rights.
• If you report an AI response (via the "Report AI response" button), you consent to share a snapshot of that session — the full transcript and the AI feedback — with our review team for safety analysis. Reviewers work from this snapshot only; they do not have live access to your sessions. You can withdraw consent at any time, which deletes the report and its snapshot.

5. Known limitations

AI-generated content can be inaccurate or misleading. Known failure modes include:

• Hallucination — the AI may generate confident but clinically incorrect statements.
• Fabricated clinical content — patient history, symptoms, or test results are fictional and may not reflect real-world presentations.
• Persona drift — the patient's behaviour may shift inconsistently during a long session.
• Bias — language models can reflect biases present in their training data; patient personas may not always represent real-world diversity accurately.

Always apply your professional judgment. Never treat AI-generated content as clinical ground truth.

6. What Practicly must NOT be used for

• Real patient consultations or clinical decision support
• Diagnostic decisions or treatment selection for actual patients
• Board certification, credentialing, or any formal assessment
• Any safety-critical healthcare application

Practicly is a training tool only.

7. Your rights

• Human review: You may request a human review of any AI-generated feedback by contacting us at [email protected].
• Complaints: Flag a specific concerning AI response with the in-product "Report AI response" button (which shares a consented session snapshot for review — see Section 4), or email [email protected].
• GDPR rights: Access, rectification, erasure, portability — see our Privacy Policy.

8. How we govern this

• A Data Protection Impact Assessment (DPIA) covering the STT/LLM/TTS pipeline is on file.
• We maintain a sub-processor list of all AI providers and data processors.
• Our governance follows EU AI Act Art. 50 transparency requirements for limited-risk AI systems.
• Material changes to AI models, prompts, or scoring logic are subject to an internal review gate before deployment.

Questions? Contact [email protected] or our AI safety address [email protected].

This document is the AI system disclosure required under Article 50 of the EU AI Act and is in effect as of 15.09.2026.