Privacy Policy & Stateless Data Security
Last Updated: October 2026. OriginSift is built on an ephemeral, stateless processing architecture. Your documents, photos, audio clips, and video recordings are processed in volatile memory and never stored on disk or used to train AI models.
Core Data Protection Commitments
No AI Model Training
Your uploads are never used to train, fine-tune, or evaluate public or private AI models.
Stateless In-Memory Processing
Files are processed strictly in volatile system memory during analysis and discarded upon completion of the request.
No Account Required
You can use the full inspection suite without submitting an email, phone number, or creating an account.
Encrypted in Transit (TLS 1.3)
All communication between your device and our processing nodes is encrypted with HTTPS / TLS encryption.
1. Stateless Data Processing Architecture
When you upload or paste an image, audio clip, video file, or text document into the OriginSift Provenance Workbench, our inspection pipeline adheres to strict volatile-memory handling:
1.1 Ephemeral Memory Lifecycle
Uploaded byte streams are read directly into volatile RAM buffers for the duration of the HTTP request. When native container decoders (such as C2PA manifest verification or FFmpeg audio/video stream demuxing) require a file descriptor, short-lived operating system temporary files are unlinked immediately inside a guaranteed finally block before the HTTP response returns.
1.2 Zero Persistent Storage of User Media
Once the inspection JSON response is computed and returned to your browser, the uploaded content is permanently discarded from server memory. OriginSift does not archive user uploads to databases, local disks, or cloud object buckets.
1.3 Standard Operational & Rate-Limiting Logs
To protect the service against automated denial-of-service floods, our Nginx reverse proxy maintains standard ephemeral access logs (IP address, request timestamp, HTTP method, and status code) and rate-limiting counters, without logging the contents of uploaded files or submitted text.
2. Our Commitment Against AI Training & Model Harvesting
Unlike many commercial generative AI platforms that reserve broad licenses over user inputs:
2.1 Strict Prohibition on Model Training
We never use submitted text, uploaded photographs, voice recordings, video clips, or Word/PDF documents to train, retrain, RLHF-align, or fine-tune our own machine learning models or any third-party foundation models.
2.2 Safe for Editorial, Academic & Investigative Workflows
Newsrooms, fact-checkers, educators, and researchers can inspect unpublished drafts and sensitive media without risking data leakage into public training corpora. Learn more about our inspection architecture on the About OriginSift & Methodology page.
3. Browser & Client-Side Privacy Safeguards
Where technically possible, OriginSift executes privacy-sensitive operations directly inside your browser:
3.1 Client-Side PDF Report Generation
When you export an inspection summary as a PDF certificate, the document is rendered locally inside your browser using client-side JavaScript (jsPDF). Your media preview and report contents are not uploaded to a third-party PDF rendering service.
3.2 Analytics & Minimal Telemetry
We use standard aggregate page-view analytics (Google Analytics 4) solely to understand overall traffic across public routes such as our Field Guides & Blog. No uploaded file names, text drafts, or inspection results are ever transmitted to analytics services.
4. Information We Do NOT Collect
OriginSift is engineered around data minimization principles:
- We do not require user registration, passwords, or personal profiles to run media authenticity or provenance checks.
- We do not sell, rent, or trade user data to third-party data brokers, ad networks, or marketing platforms.
- We do not store persistent copies of images, videos, audio recordings, or documents submitted for inspection.
5. Global Privacy Rights (GDPR, UK GDPR & CCPA)
Because OriginSift does not require user accounts or retain uploaded media files, we do not store personal media profiles on our servers. Under the General Data Protection Regulation (GDPR) and California Consumer Privacy Act (CCPA), users retain full statutory rights over their personal information. Please also review our Terms of Service & Evidentiary Limitations.
6. Contact & Data Protection Inquiries
If you have questions regarding our stateless architecture, C2PA verification pipeline, or privacy practices, please reach out to our engineering team through Atanico Digital Solutions.