Union des Photographes Professionnels joins CAI

The Content Authenticity Initiative ( CAI ) is an organization founded in November 2019 by Adobe, the New York Times and Twitter, which today has over 3,700 members worldwide, including major industry players, including international social networking platforms, generative AI, media, major Tech companies, among the largest camera manufacturers, individuals from civil society, non-profit organizations . Its aim is to combat misinformation and enable the identification of the provenance of content disseminated on the internet. and much more besides.

To this end, CAI is a co-creator of the Coalition for Content Provenance and Authenticity (C2PA), an international organization working to implement open source technical standards: Content Credentials. This standardized format of provenance metadata is intended to act as a "digital label". CAI's mission is to promote the widespread adoption of these standards, so that transparency and content labeling become commonplace. Content credentials are already used by Google, Meta, LinkeIn, Amazon, TikTok, OpenAI...

C2PA metadata structure in a file with several manifests generated during image registration, editing and publication

The procedures proposed by CAI and C2PA make it possible to combat misinformation [ 10 ] [ 11 ] on a large scale, by using a set of additional data (metadata) containing details of the provenance of information displayed on a digital device. This information can be, for example, a photo, video, sound or text file. C2PA metadata for this information can include, among other things, the publisher of the information, the device used to record the information, and the location and time of the recording or editing steps that modified the information. To ensure that C2PA metadata cannot be altered without anyone noticing, it is secured with hash codes andcertified digital signatures . The same applies to the main content of the information, such as an image or text. A hash code of this data is stored in the C2PA metadata section, and then, as part of this metadata, secured with the digital signature.

Securing metadata and main content with certified signatures enables users to reliably identify the source of a file they are viewing. If C2PA metadata identifies, for example, a certain TV channel as the publisher of a file, it is highly unlikely that the file comes from any other source.

Files containing C2PA-compliant metadata that are copied from a publisher's website, then published unmodified on social networks (or elsewhere), always retain the full set of tamper-proof provenance information. Users viewing such content on social networks can examine such a file with an online tool offered by CAI [ 12 ] or, where applicable, with C2PA-compliant inspection tools offered by the social network site. Standard-compliant tools will detect whether unauthorized modifications have been made to the file or metadata. If no modifications have been made, the user can trust the metadata and main content to be exactly as published.

The methods proposed by CAI and C2PA cannot determine whether content is "true", i.e. whether it contains authentic information that faithfully reflects reality. Instead, C2PA-compliant metadata only provides reliable information about the origin of a piece of information. Users' confidence in this information depends solely on their confidence in its source.