The National Security Agency‘s Artificial Intelligence Security Center, in collaboration with international agency partners, has released new guidance discussing how content credentials could help increase the transparency of images, audio, video and other media amid the adoption of generative AI tools. However, the organization subsequently removed the press release announcement from their website.
NSA said Wednesday the cybersecurity information sheet, or CSI, which is still online, provides information based on current threats and techniques and explains how the Coalition for Content Provenance and Authenticity, or C2PA, created an open specification for providing digital media provenance through content credentials.
The CSI also presents questions that organizations should consider when preparing for content credentials implementation and recommends practices to ensure the preservation of unaltered metadata throughout the media lifecycle.
“Bolstering trust through transparency in multimedia has never been more critical. Secure and widespread adoption of content provenance standards is a ‘must’ to prepare us for the new reality where AI is everywhere, and seeing is no longer believing,” said Candice Rockell Gerstner, applied research mathematician at NSA.
What Are Content Credentials?
According to NSA, content credentials are cryptographically signed metadata that enable creators to directly incorporate into media content information about themselves and their creative process.
The metadata can be added to the media content at creation on hardware or during export from software and can be made more durable through the addition of a digital watermark to the media and implementation of a fingerprint matching system.
NSA developed the document in partnership with the Australian Signals Directorate’s Australian Cyber Security Centre, the Canadian Centre for Cyber Security and the U.K. National Cyber Security Centre.