SafeImage NZ
Schools and NGOs publish images of children constantly with almost no systematic safeguarding process behind it. SafeImage NZ analyses a photo before it goes out and explains, in plain language, exactly what it reveals — a school crest, GPS metadata, a visible timetable — then handles consent tracking, an audit trail, and safe-version generation in one workflow.
Key Features
- Threat Simulator — plain-language AI risk analysis
- Risk scoring (0–100) per image
- Consent tracking with missing/expired flags
- Audit trail of uploads and approvals
- Safe image generation — face blur, stripped metadata, smart crop
- Downloadable safeguarding report
Technologies
- Next.jsReactTailwind CSSAnthropic APICanvas API
Challenges
Turn a compliance checkbox most schools and NGOs treat as a formality into something that actually catches real risk, without asking staff to become privacy experts.
Solution
An AI Threat Simulator that explains exactly what a bad actor could learn from a photo, paired with consent tracking, an audit trail, and one-click safe-image export — all processed client-side, so images are never sent to or stored on a server.
Results
A working prototype covering the full loop — upload, AI analysis, consent check, safe export, report — built and demoed solo in two days at the AI Hackathon Festival, after two teammates had to step back before the event.
Gallery