"Cryptography May Offer a Solution to the Massive AI-Labeling Problem"

The European Union (EU) will soon require some technology platforms to label their Artificial Intelligence (AI)-generated images, audio, and videos with "prominent markings" showing their synthetic origins. In addition, the White House wants major AI companies to disclose when their content was created using AI. However, identifying material created by AI is a significant technical challenge. According to researchers, the best available options, such as AI-powered detection tools and watermarking, are inconsistent, temporary, and occasionally inaccurate. C2PA is another approach that has recently garnered much interest. It is a relatively new open source Internet protocol that uses cryptography to encode information about the origins of a piece of content, or "provenance" information. The creators of C2PA compare the protocol to a nutrition label, except that it reveals the origin of the content and who or what created it. The project, which is part of the nonprofit Joint Development Foundation, was initiated by Adobe, Arm, Intel, Microsoft, and Truepic, who formed the Coalition for Content Provenance and Authenticity (from which the protocol gets its name). This article continues to discuss what C2PA is and how it is being used.

MIT Technology Review reports "Cryptography May Offer a Solution to the Massive AI-Labeling Problem"