Fox Releases Verify, An Open-Source AI Content Solution
Over the past year media companies have been wrestling with the explosive growth in AI, particularly in the form of Large Language Models (LLMs) that have been trained on vast reams of their online content and now are using that learning to automatically generate new content — usually without the original creator providing permission or receiving any compensation. Broadcasters and publishers also worry about their trusted news brands being associated with fake, or simply out-of-context, images appearing in stories generated by AI bots.
Some publishers have simply blocked AI crawlers from accessing their websites, while others have sought legal recourse, such as the New York Times’ ongoing lawsuit against OpenAI. Some publishers are seeking multiyear deals to license their copyrighted work to AI companies in order to train their models.
Fox Corp. believes it has developed a solution to both of AI’s major intellectual property problems with Verify, a technical protocol that it is now making available on an open-source basis. Verify allows media companies to register their content and grant usage rights to AI platforms, such as attribution and/or compensation, through a “Verified Access Point” and “smart contracts” executed via blockchain. Verify also allows consumers to verify the authenticity of content via Polygon PoS, an open source blockchain system.
Fox is making Verify, which was previewed by Tubi Media Group CEO Paul Cheesbrough in a TVNewsCheck webinar last month, publicly available starting today. The company has also released a beta version of the first Verify application, the “Verify Tool,” a web-based solution that gives consumers the ability to authenticate ownership of images and article links they find on the internet claiming to be from trusted sources. Fox has its own application of the Verify Tool, which it plans to promote on Fox News, and will be found at verify.fox.
While Fox wants to use Verify to create new commercial opportunities by licensing its content to AI platforms, it isn’t looking to monetize the protocol directly, says Fox Corp. CTO Melody Hildebrandt. By making the code available on an open-source basis, Hildebrandt says, Fox hopes to create a “flourishing ecosystem” around the protocol with broad adoption among media companies and developers creating their own third-party applications based on it, such as browser plug-ins that might automatically verify news content. She adds that Fox has received positive feedback about the technology in early conversations with both media companies and AI platforms.
“Our view is we could actually come out with an architecture that would establish the legal and verified mechanism by which our content could be used for these downstream applications, and that we could establish that architecture rather than having Big Tech impose it upon us,” Hildebrandt says.
Verify was developed in-house by the Fox Technology team over the past year in collaboration with Polygon Labs, leveraging blockchain technology infrastructure that Fox had already created for content syndication applications. It uses Polygon Labs’ PoS open source blockchain, advanced searching, and scalable ZK (zero knowledge) technology.
The Verify protocol establishes the origin and history of original journalism by cryptographically signing individual pieces of content on the blockchain, which in turn allows the content to be independently verified by consumers as having originated from the expected source using the Verify tool. Fox launched a beta version of Verify on Aug. 23, 2023, to coincide with the first Fox News GOP debate. Since then, online content from Fox News, Fox Business, Fox Sports, and most recently, Fox Television Stations, has been run through the Verify system, totaling around 84,000 signatures for all content including both articles and images. Fox is not currently using Verify in its broadcast feeds.
The Verify pipeline, which connects to Fox’s content management system (CMS), takes about a minute to “fingerprint” an article. Each story that runs through Verify creates a “content graph,” a tree structure that includes the headline, embedded images, videos and the text, all as part of an integrated graph. Hildebrandt says that Verify should be easy for other media companies to adopt in their online publishing workflows.
“The code is all being open-sourced, so a publisher could pick up this code, essentially integrate it into their CMS and be off and running, and not need any additional care and feeding,” she says.
The Verify Tool uses the information contained in the Verify content graph to allow consumers to check the authenticity of an online story via the web portal. A consumer can right-click and save an image, or simply screenshot the entire page, and then upload it to Verify Tool. The system will then indicate whether the image actually came from the identified source. While “deep fakes” and other AI-manipulated images have worried broadcasters for years, a more recent problem is the out-of-context use of real images in social media, such as a picture from 2015 appearing in a purported 2024 news story describing the conflict in Gaza.
“This is a brand protection exercise as well,” Hildebrandt says. “In this age of AI proliferation of content, our brand is more important than ever. And consumers are going to rely on the brands they trust to be the interface between them and some of these pieces of information. We thought it was important to get ahead of that consumer problem.”
Fox is continuing to improve Verify and a second release is targeted for later this year. Fox also expects to announce content partners for the technology in the next few months.