A simple system: Take a picture. A unique fingerprint goes on a public blockchain. Anyone can verify it came from a real camera, not AI.
Modern AI can create photorealistic images that are indistinguishable from real photographs. How do you know what's real?
Social media platforms strip authentication data when you share images. Existing solutions like C2PA lose 95% of their metadata in real-world use.
Amid the sea of content both real and fake, journalists, photographers, and the public need a way to prove which images came from real cameras.
Your camera captures the image and immediately creates a unique "fingerprint" (called a hash) of the raw sensor data. This happens automatically in the background.
Think of it like a DNA sample from the image—unique and unforgeable.
That fingerprint is sent to a blockchain operated by journalism organizations, universities, and archives. They store it permanently in a public database.
No one stores your image—just the fingerprint. Your photo stays private.
When someone sees your image online, they can check its fingerprint against the blockchain. If it matches, they know it came from a real camera at a specific time.
Works even if the image was converted, cropped, or shared on social media.
The Birthmark Standard is designed to be incapable of betraying you—not through policy, but through architecture.
Authentication creates two separate data structures sent to different parties that are never combined. Manufacturers validate cameras without seeing what you photographed. The blockchain stores image hashes without knowing which specific camera authenticated them.
Learn more →Your device is never a "party of one." Every camera randomly selects from assigned key tables shared by over 1,000 devices. Individual tracking is mathematically infeasible without manufacturer cooperation.
Learn more →No single server can forge a blockchain record. Every entry requires independent validation from two geographically separated submission servers. A single compromised operator cannot create fraudulent authentication.
Learn more →Unlike systems that hide data in file headers, Birthmark identifies the "DNA" of the pixels themselves. Your authentication survives re-compression, format conversion, social media sharing, and even screenshots.
Learn more →We've built and validated a complete working system using Raspberry Pi hardware:
Economics: Operating a blockchain node costs $200-350/year even at 1 million images per day, making it sustainable for journalism institutions.
We're producing a complete demonstration video showing the end-to-end workflow: capturing an image, submitting it for validation, and verifying it on the blockchain.
The same principles that authenticate photos can extend to other media types:
Hash individual frames or frame sequences to prove video came from real cameras. Detect deepfake insertions or frame manipulation.
Target: Phase 3
Validate audio recordings from microphone hardware. Detect AI-generated voices or manipulated speech.
Target: Phase 3
Prove screenshots came from specific devices at specific times. Useful for documenting online content before deletion.
Target: Phase 4
All code is public. No corporate secrets or proprietary control.
Only image fingerprints are stored, never the images themselves. Photographers stay anonymous.
No single company controls the network. Multiple institutions operate blockchain nodes.
No gas fees, no transaction costs. Journalism organizations donate hosting as a public service.
C2PA provides rich metadata. Birthmark provides backup when metadata gets stripped.
Works even when images are shared on social media, converted to different formats, or cropped.
Join our pilot testing program. Help us understand authentication needs for your workflow.
Review the architecture, contribute code, or help with security audits.
Explore hardware integration opportunities and partnership possibilities.