AI, Educational Use and Clinical Safety Policy

2. No clinical use; no patient information

Do not use SurgeonAI to diagnose, treat, triage, prescribe for, or make decisions about a real patient. Do not enter PHI, patient records, identifiable case histories, operative footage, medical images, or any other patient-identifying information. The Service is not an emergency service.

If you encounter a real clinical concern, follow your local protocols and seek appropriate senior, specialist, or emergency support. Do not wait for, or rely upon, AI output.

3. Understanding AI output

AI-generated content can be wrong, incomplete, outdated, overly confident, or insensitive to local practice, patient context, or changing evidence. It may mishear spoken answers, misunderstand a question, or classify an answer differently from a human examiner.

Treat every output as a prompt for learning and verification. Check important information against current, authoritative sources, applicable guidelines, and appropriate human supervision. Report a concern using safety@surgeon.ai.

4. Scores, safety labels, and readiness bands

Scores, “safe/unsafe” classifications, critical-fail flags, percentiles, competency profiles, and readiness bands are educational indicators generated from limited evidence. They are not guarantees of competence or patient safety, official results, certification, or a substitute for a human assessment.

SurgeonAI does not permit these outputs to be the sole basis for a high-impact decision about employment, admission, promotion, training progression, credentialing, privileges, discipline, or fitness to practise. Institutions and educators must ensure meaningful human review, context, transparency, an opportunity to challenge material inaccuracies, and a fair process before making any consequential decision.

5. Voice, video, and recording

Only record yourself and people who have agreed to be recorded where consent is required. Keep recordings professional and free of patient information. We do not use voice recordings for biometric identification or voiceprint profiling.

You may have controls to delete recordings or ask an institution to remove them, subject to lawful retention needs and its configuration. See the Privacy Notice for intended retention periods.

6. Content quality and educator controls

We seek to use original educational content mapped to public curriculum blueprints and to subject AI-assisted authoring to human approval. We do not claim that content is exhaustive, current in every jurisdiction, or equivalent to official examination material.

Where the Service says that an educator has reviewed a skills-video result, a human educator must actually have reviewed it. AI may help structure a draft or scorecard, but it must not be represented as human review.

7. Fairness, safety, and feedback

We aim to identify and reduce avoidable bias and unsafe behaviour through evaluation, monitoring, feedback channels, version control, and human review. Because no such process is perfect, we encourage users to report:

• a clinically unsafe or materially inaccurate answer;
• discriminatory, offensive, or biased content;
• a transcription or scoring error;
• a suspected breach of privacy or use of patient information; or
• an attempt to use the Service to falsify experience, assessments, or credentials.

We may investigate, correct, limit, or remove affected content or features. If a serious issue could affect user safety, we may suspend the relevant feature while we investigate.

8. Data and model-improvement commitment

We do not use identifiable user recordings, transcripts, reflections, logbook entries, portfolio data, or assessment content to train a general-purpose AI model without clear, separate opt-in consent. We may use de-identified, aggregated metrics to evaluate and improve the Service where permitted by law. See the Privacy Notice for full details.

9. Your responsibilities

Use the Service honestly and professionally. Do not attempt to manipulate assessment tools, evade safeguards, impersonate another person, submit work performed by another person as your own, or present SurgeonAI output as an official score or credential. You remain responsible for your clinical practice, learning, professional conduct, and compliance with local rules.