Safe AI Policy

Ensuring responsible and ethical integration of artificial intelligence in clinical workflows

1. Purpose and Scope

This policy sets out the principles and operational guidelines for the safe, responsible, and ethical use of artificial intelligence (AI) technologies within the veterinary practice. It applies to all AI systems in use, including but not limited to:

Vision-based AI systems (e.g. for wound detection, lameness scoring, image interpretation)

Boop-integrated AI functionalities

AI-enabled diagnostic tools

AI scribe or clinical documentation assistants

2. Guiding Principles for AI Use

Transparency: AI use must be clearly disclosed to staff and clients when it affects care, diagnoses, or documentation.

Human Oversight: AI tools must support—not replace—clinical judgment. All outputs must be reviewed by licensed veterinary professionals before action.

Safety and Reliability: Only validated, well-tested, and up-to-date AI tools should be used. All systems must have fallback procedures in case of malfunction.

Privacy and Data Security: All AI systems must comply with data protection regulations (e.g. GDPR) and maintain the confidentiality of both client and pet data.

Fairness and Non-Discrimination: AI systems should be evaluated regularly for biases, especially regarding image interpretation across species, breeds, and presentations.

3. Vision Technology

Approved Use Cases

Skin lesion identification

Radiograph and ultrasound interpretation

Movement/gait analysis

Safe Usage Guidelines

Vision AI tools must be trained on diverse, species-appropriate datasets.

Output should include a confidence score and human-readable summary.

All images and AI results should be stored securely in the pet’s medical record with a timestamp and reviewer’s name.

4. Boop®-Integrated AI Systems

Safe Usage Guidelines

All AI prompts or actions initiated via Boop must be logged and traceable.

Any AI-generated recommendation (e.g. treatment plans) must be reviewed by a registered veterinary surgeon.

Feedback loops should be implemented to continuously improve AI suggestions using clinician input.

5. AI-Enabled Diagnostic Tools

Example Applications

Vetscan Imagyst™ for digital microscopy and cytology

Differential diagnosis generation

Lab result interpretation

Predictive analytics for disease risk

Safe Usage Standards

Always cross-check AI-generated diagnoses with traditional methods.

Diagnostic tools must be reviewed and approved by the Clinical Director before deployment.

Regular audit of AI outputs vs. actual outcomes to ensure safety and accuracy.

6. AI Scribe and Documentation Assistants

Usage Protocol

Used for transcribing consultations, summarising medical history, and drafting SOAP notes.

All documentation must be reviewed and signed off by the attending veterinary surgeon before it is finalised.

Audio data and transcripts must be encrypted and stored securely.

7. Training and Compliance

All staff must receive annual training on AI use, limitations, and safety.

Internal audits on AI use will be conducted quarterly.

Non-compliance with the AI policy may result in restricted system access or disciplinary action.

8. Incident Reporting and Review

Any suspected AI-related error, bias, or malfunction must be reported immediately to the Clinical Director.

All incidents will be documented and reviewed with appropriate remediation plans.

9. Review and Updates

This policy will be reviewed biannually and updated to reflect technological advances, regulatory changes, and clinical feedback.