Proof of Capture: The $100 Open-Source Answer to Apple's Image Authenticity

María Benavente and Alex Hornstein have developed "Proof of Capture," an open-source camera system designed to cryptographically verify a photograph's authenticity at the exact moment it is taken. The project offers a transparent, hardware-based alternative to proprietary systems like Apple's Reference Image, aiming to provide a verifiable chain of trust for digital imagery.

The system is built on affordable hardware, utilizing a Raspberry Pi Zero, a display board, and an ATECC608 cryptographic chip. To ensure the integrity of the process, the ATECC608 chip manages the signing process such that the private key never leaves the hardware; the chip is designed to lock itself if any physical tampering is detected. According to the project's documentation, the entire system is open source and can be constructed for less than $100.

The Shift Toward Provenance

This development comes as the tech industry pivots from "detection" to "provenance." While traditional AI detection attempts to spot fakes after they are created, provenance focuses on proving an image is real from the source. While major manufacturers like Nikon and Sony have adopted the C2PA industry standard, Apple has introduced its own Reference Image system, which relies on Private Cloud Compute as its root of trust.

María Benavente argues that relying on detection is a futile effort. "Detection is a losing race: every improvement in the detector is training signal for the next generator," Benavente stated. "Our approach flips the problem: instead of trying to detect what's fake after the fact, prove what's real at the moment of capture."

Solving the Metadata Problem

Proof of Capture distinguishes itself by avoiding reliance on corporate cloud infrastructure or fragile metadata. Most authenticity proofs rely on EXIF data or sidecar files, which are typically stripped away when a photo is uploaded to social media or sent via messaging apps.

To solve this, the system employs steganography. It uses a frequency-domain watermark—specifically combining Discrete Wavelet Transform (DWT) and Discrete Cosine Transform (DCT)—to embed a signed perceptual hash (pHash) directly into the image pixels. Because the proof is woven into the visual data itself rather than attached as a label, the authenticity signature can survive the heavy compression typical of platforms like WhatsApp and can still detect if the content has been edited.

The Future of Digital Trust

As AI-generated imagery becomes increasingly indistinguishable from reality, the existence of a decentralized, low-cost verification tool provides a blueprint for non-corporate digital trust. By moving the root of trust from a cloud server to a physical, open-source chip, the project demonstrates that authenticity does not require a subscription or a proprietary ecosystem.

Observers will now be watching to see if this steganographic approach can be scaled to mainstream camera firmware or if the industry will continue to consolidate around cloud-based proprietary standards.