How PicFetch AI safeguards wedding couples, guests, and photography studios with zero selfie retention, irreversible mathematical vectors, and strict statutory compliance.
As a professional photographer in India, your reputation is your business. Under the Digital Personal Data Protection Act, 2023, businesses face statutory penalties of up to ₹250 Crore for failing to implement reasonable security safeguards when handling biometric data. PicFetch AI provides the technical and statutory shield you need.
You act as the event Data Fiduciary; PicFetch AI operates as your contractually bound Data Processor. All processing is legally categorized and compliant.
If any wedding attendee or VIP requests to be removed, their face embedding is deleted in seconds with an immutable statutory audit receipt.
Automated 60-day purging prevents old albums from lingering as unmonitored liabilities, keeping your storage lean and your compliance spotless.
Every design decision in our platform prioritizes user privacy, data minimization, and regulatory rigor.
When a guest takes a selfie to find their photos, the image is decoded strictly in temporary server memory (RAM) to calculate mathematical feature coordinates and is immediately discarded. It is never written to disk (/tmp), never saved in PostgreSQL, and never uploaded to cloud storage buckets.
Faces in studio photos are converted into 512-dimensional numerical vectors using AuraFace ArcFace deep learning. These numbers represent cosine distances, not images. It is mathematically impossible to reconstruct a human face or regenerate a photograph from an embedding.
Every facial query enforces strict database filtering (`WHERE Face.event_id = :event_id`). Vectors from Wedding A can never match, search, or leak into Wedding B. We do not maintain a global biometric registry and never track individuals across different events.
To prevent guests from holding up another person's photo on an iPad or phone screen to steal their pictures, our browser camera interface prompts an interactive motion challenge ('Blink naturally' or 'Turn head slightly') and measures inter-frame optical delta before taking the shot.
In compliance with India's DPDP Act Section 12, any guest can instantly erase their face data (`/erase`). Solo photos are automatically hidden from public view, while in group photos only that individual's vector is excised—preserving memories for other guests without ever matching that person again.
Private event data is not hoarded forever. Daily background maintenance workers automatically delete original high-resolution photos, previews, thumbnails, and pgvector embeddings once an event reaches its retention period, leaving zero orphaned biometric liability.
When a guest matches their photos, the server issues a time-limited HMAC-SHA256 signed bearer token specifying the exact matched photo IDs. A guest cannot alter URL query parameters, guess photo UUIDs, or bulk download un-matched photos from the event.
Direct PostgREST queries from browser or mobile clients are blocked by default via PostgreSQL Row-Level Security (RLS) on all public tables. All interactions must route through authenticated, rate-limited FastAPI backend REST endpoints.
Direct, transparent responses to the most common privacy and data security concerns.
Copy and paste these pre-approved privacy disclosures directly into your client proposals or WhatsApp messages.