As the leading global news agency, Reuters works relentlessly to bring news from the source and provide readers and viewers with unbiased and reliable information from the world around us. Every day, our journalists capture the first draft of history, without bias or agenda, to inform billions of people.
Photojournalism plays a crucial role in documenting that history. Yet, with generative AI technology now available to anyone with internet access, it has become much easier to create visuals that can deceive or misinform. To make sure we can protect the trust we have built up over decades, we're always exploring innovative technologies. This is one big step forward.
Many photojournalists worldwide rely on Canon technologies. As a leading global camera manufacturer, Canon understands the role images play in society and recognises the importance of preserving image authenticity. Working as part of the Content Authenticity Initiative (CAI) alliance, Canon wants to take meaningful measures to protect image outputs that serve the news community worldwide.
In our proof of concept, a picture taken by our photojournalist is digitally signed by a Canon camera using a device-specific key at the point of capture and attached to the image, forming the first version of the picture. The authenticated picture's pixels, GPS and other metadata, is sent to the Reuters system directly from the camera and registered onto a public blockchain as well as preserved on two cryptographic archives that use the Filecoin and Storj protocols.
A news consumer can check the validity of a picture by comparing its hash to the value in the public ledger system. The values should be the same, irrespective of where the consumer retrieves the picture. If they are not, the image has changed since Reuters published the picture, breaking the chain.
Collaborating with the teams at Canon and Starling Lab, we tested the technology while covering the war in Ukraine, a story in which there has been significant mis- and dis-information. Reuters photographer Violeta Santos Moura captured pictures from the frontline using a prototype Canon camera which digitally signed every picture taken.
The scope of our proof of concept was two-fold:
1. To preserve the chain of integrity of a picture taken by a Reuters photographer from camera through to the point of publication.
2. To test the integration of authentication technology on our picture desk workflows.
Generating the digital signature including hash value calculation, signing and injection into picture data required additional processing time on the test camera, increasing the difficulty for the photographer to capture the best quality pictures under the added pressure of operating in a conflict zone.
Every second counts in our picture editing and delivery pipeline, and the impact of integrating authentication workflows to our delivery speed was significant.
In our standard FotoWare ingest workflows, minor automated edits are made to the picture metadata on upload. Since any change breaks the chain of authenticated C2PA information, we needed to find a way to capture these edits to avoid breaking the C2PA provenance history.
FotoWare developed a procedure to re-process the asset for C2PA signage through the Starling Lab system to maintain content authenticity for the proof of concept. With this important reference implementation, work can accelerate to ensure that all digital asset management (DAM) systems used by our customers can leverage the C2PA's open source standard end-to-end.
Given the complexity and degree of novelty of the technology applied in this proof of concept, we would like to hear from our customers about its potential application in their content ingest, editing and publication workflows.
Digital signature is a mathematical method for verifying the authenticity of digital contents. Using a hash function, a piece of digital information is converted into a hash value: a unique digital fingerprint that is basically a large number. Any two pieces of content that differ, no matter how small, will generate different hash values, making it very easy to detect if the information has changed.
A hash value is calculated from entire or parts of a target digital content and digitally signed using a private key for digital signing. The signature data can be used by a recipient to verify the origin of the digital content.
In the verification process, the signature is verified using a public key which was made along with the private key. Only a paired public key can verify the signature so it gives the recipient reason to believe the content was made by a certain person.
Blockchains are composed of blocks, whereby each block contains three pieces of information: data (or a reference to it), a hash value calculated from that data, and the previous block's hash value. Each new version of the data is included in a new block, and the "chain" of preceding values makes it possible to track every modification back to the original material.
Blocks are stored in public directories called ledgers. Public ledgers are distributed across multiple sources to prevent tampering. A malicious third party that wishes to tamper with the chain of blocks would have to do it across multiple systems that they would not have full access to, at the same time. This is extremely difficult to do. Finally, multiple publicly distributed ledgers can be used for the same data content for added security.
The Coalition for Content Provenance and Authenticity (C2PA) is a Joint Development Foundation project to collectively build an end-to-end open technical standard to provide publishers, creators, and consumers with opt-in, flexible ways to understand the authenticity and provenance of different media types.
FotoWare provides the digital asset management (DAM) system used as a hub for Starling Lab's implementation of the C2PA specification. Media with provenance information uploaded to the DAM is automatically digitally signed to allow consumers to verify content credentials via the DAM web interface. FotoWare discovered the value of establishing trust in assets modified within the DAM by users or the system itself so a procedure was developed to re-process the C2PA signature through the Starling Lab system to maintain content authenticity.
The C2PA standard and its media signing tools are used throughout the system, to bundle and embed provenance information from various stages directly into the authenticated pictures in our final distribution.