By Felix Chen
At our June workshop, organized jointly by Princeton’s Center for Information Technology Policy (CITP) and NYU Journalism, participants shared a desire for a practical, short-form resource to help newsrooms understand the current landscape of authentication and verification tools for addressing AI-generated content. To complement our full report, we have translated the workshop discussions into a short guide on considerations for newsroom workflows.
The newsroom guide is now available online.
This guide emphasizes one of the key sentiments coming out of the workshop – the need to conceptualize tools not as definitive classifiers of “real” or “fake,” but rather as components in a broader workflow, in which different tools can be used at different stages to provide, in aggregate, greater confidence in the authenticity or veracity of any particular piece of content.
We distinguish between two very closely related purposes for tools – authentication and verification – and use that distinction to categorize tools for different use cases. Authentication makes a determination limited to the provenance of an item. Verification evaluates whether the content of that media depicts what it claims to depict. This distinction is important because satisfying one objective does not necessarily imply the other. For example, a photo may be authentic—with accurate information on when and where it was captured – but that does not mean that the photo supports claims accompanying it. These objectives are separate and both require scrutiny; one does not substitute for the other.
This guide draws together a list of tools that could be used to address these two objectives across different media types – images, audio, video, and documents – to provide a starting point for thinking about different sets of tools for authentication and verification and how they might fit into a broader newsroom workflow. We draw extensively from Bellingcat’s Open Source Toolkit, the Global Investigative Journalism Network (GIJN), Tom Vaillant’s OSINT Tool Database, reporting from Deutsche Welle (DW), and conversations with workshop participants to develop this list.
This list is meant to serve as a starting point rather than an endorsement of these specific tools. CITP has not independently tested every tool.
Further Reading:
- Full CITP report: Holding the Line: Authentication, Verification, and the Fight for Facts in the AI Age
Press:
- From Nieman Lab: “AI authentication tools are built without adequate journalist input, new report finds”
- From Tech Policy Press: “What It Now Takes to Establish a Fact“
Felix Chen is a researcher interested in examining societal impacts of artificial intelligence, technology-enabled access to justice, and digital public infrastructure. Previously, as a volunteer for the Massachusetts Small Claims Advisory Service, Chen developed technical tools aimed at improving public access to legal information.
Leave a Reply