SAFEGUARDING POLICY

1. Our Commitment

Spark Tradie is committed to the safety, health, and wellbeing of all users, particularly young apprentices. The Australian trades industry has one of the highest rates of mental health challenges and suicide among young workers. We believe technology can play a vital role in early intervention and support.

This Safeguarding Policy outlines how our AI-powered wellbeing system detects risk, responds to crisis situations, and connects at-risk individuals with appropriate support — while respecting privacy and building trust.

2. Scope

This policy applies to all interactions with Spark AI, our wellbeing coaching feature. It covers:

• All text-based conversations with Spark AI
• Voice conversations with Spark AI (when voice features are enabled)
• Risk assessment and classification processes
• Alert protocols for RTO wellbeing coordinators
• Crisis response procedures
• Data handling for wellbeing information

3. Three-Tier Risk Assessment

Every message sent to Spark AI is assessed for risk indicators using our safeguarding system. Messages are classified into three tiers:

4. How Risk Detection Works

Our safeguarding system uses two layers of assessment:

4.1 Pattern-Based Detection

User messages are analysed against established risk-indicator patterns, including keywords and phrases associated with suicidal ideation, self-harm, abuse, and extreme distress. These patterns are regularly updated based on mental health research and Australian crisis service guidelines.

4.2 Contextual AI Assessment

The AI coaching model (powered by Anthropic Claude) is instructed to recognise escalating distress within conversation context. This provides a second layer that catches nuanced expressions of risk that keyword patterns alone may miss.

4.3 Limitations

No automated system can detect all instances of risk. Our safeguarding system is designed as a safety net, not a replacement for human intervention. Users are always encouraged to contact crisis services directly, and crisis resources are visible throughout the wellbeing experience.

5. RTO Wellbeing Alerts

5.1 What Alerts Contain

• Risk tier classification (HIGH or MEDIUM)
• AI-generated summary of the concern (1-2 sentences)
• Timestamp of the triggering message
• Session ID for reference
• User's name and RTO association

5.2 What Alerts Do NOT Contain

• Full conversation transcripts
• Historical wellbeing conversation data
• Other personal information beyond what is necessary for outreach

5.3 Alert Handling by RTOs

RTO wellbeing coordinators who receive alerts must:

• Acknowledge receipt of the alert within the RTO Portal
• Initiate contact with the apprentice within one business hour for HIGH-risk alerts
• Document the outcome and any follow-up actions taken
• Mark the alert as "Resolved" with resolution notes once appropriate support has been provided
• Comply with their own organisation's safeguarding and mandatory reporting obligations

5.4 Alert Retention

Wellbeing alert records are retained for 7 years in accordance with RTO record-keeping requirements under the Standards for Registered Training Organisations 2015.

6. User Consent and Transparency

• Users are informed during onboarding that Spark AI includes a safeguarding system
• The wellbeing interface clearly states that Spark AI is not a substitute for professional help
• Crisis resources are permanently visible within the wellbeing experience
• Users are informed that HIGH-risk conversations may trigger a confidential alert to their RTO wellbeing coordinator
• Users can view their own wellbeing session history and risk tier classifications
• Users can request deletion of their wellbeing data (subject to mandatory reporting obligations)

7. Mandatory Reporting

Spark Tradie complies with all Australian mandatory reporting laws. In addition to the automated safeguarding system:

• If we become aware of a child at risk of abuse or neglect, we will report to the relevant state/territory child protection authority
• If we become aware of imminent risk to life, we may contact emergency services (000) directly
• RTO wellbeing coordinators are reminded of their own mandatory reporting obligations under their respective state/territory legislation

8. Spark AI Coaching Boundaries

Spark AI is designed as a peak-performance coaching tool. It is explicitly not a mental health service. Spark AI:

• Does: Provide motivational coaching, goal-setting support, reframing techniques, and positive mindset strategies
• Does: Encourage professional help when users express significant distress
• Does: Prominently display crisis resources at all times
• Does not: Diagnose or treat mental health conditions
• Does not: Provide medical, legal, or financial advice
• Does not: Replace professional counselling, psychology, or psychiatry
• Does not: Claim to be a therapist or counsellor

9. Data Protection for Wellbeing Information

Wellbeing data receives the highest level of protection on our Platform:

• All conversation content is encrypted at rest using AES-256 encryption
• All data is transmitted over TLS/HTTPS
• Wellbeing data is stored in the Supabase ap-southeast-2 (Sydney) region
• Access to wellbeing data is restricted via Row Level Security policies — users can only access their own data
• AI coaching responses are generated in real-time and are not stored by our AI provider (Anthropic)
• Wellbeing conversation history is automatically purged after 12 months
• Aggregate, anonymised wellbeing trend data may be used to improve the Platform — individual conversations are never used for this purpose

For full details, see our Privacy Policy.

10. Continuous Improvement

We are committed to continuously improving our safeguarding system. This includes:

• Regular review and updating of risk-detection patterns based on mental health research
• Collaboration with mental health professionals and crisis services
• Feedback from RTO wellbeing coordinators on alert quality and response outcomes
• Annual review of this policy by our safeguarding team
• Staying current with Australian safeguarding legislation and best practices

11. Reporting a Safeguarding Concern

If you have a safeguarding concern about a Spark Tradie user, or about the operation of our safeguarding system, please contact us immediately: