This system can sort real pictures from AI fakes

February 23, 2026

TL;DR

Generative AI is making it difficult to distinguish real images from fakes.
The C2PA (Coalition for Content Provenance and Authenticity) is developing a standard to embed verifiable metadata into digital content.
This metadata, similar to a 'nutrition label,' provides information about an image's origin and any modifications made.
Companies like Microsoft, Adobe, Google, and OpenAI support the C2PA standard.
Camera manufacturers like Sony and Leica are beginning to embed C2PA data, with others pledging to follow.
Image editing software like Adobe Photoshop and Lightroom can also embed C2PA credentials.
Online platforms like X and Reddit currently do not widely display C2PA metadata, hindering public verification.
Even with a standard, issues like metadata stripping (e.g., screenshots) and persistent denialism pose challenges.
Detection methods for AI-generated content are unreliable, making cryptographic labeling like C2PA a more promising approach.
The effectiveness of C2PA relies heavily on broad adoption by platforms, hardware, and software.

Continue reading the original article