Misinformation Fingerprints™ Process

NewsGuard’s Misinformation Fingerprints™ are a human- and machine-readable catalog of the top false narratives spreading online. They provide human-vetted, transparently sourced data, alerting licensees to significant false claims as they arise, detailing their provenance, and tracking how they spread across social media platforms and the open web.

Licensees include social media companies seeking to mitigate falsehoods on their platforms, and the defense and intelligence communities seeking to counter Russian, Chinese, Iranian and other information operations hostile to democracies.

The Misinformation Fingerprints™ are created using a proprietary process conducted by NewsGuard’s team of expert misinformation and disinformation researchers—aided by machine learning and other technology tools:

Narrative Identification

Narrative Evaluation

• Assessment: Once a potentially false narrative is identified, NewsGuard’s team attempts to ascertain whether the narrative is true or false. The journalistically trained analysts apply open-source research tools and techniques for evaluating narratives, establishing authoritative information to confirm or debunk each narrative, such as evidence from credible sources rated by NewsGuard, primary sources and documents, open-source public data, and subject matter experts. Narratives are only established as Misinformation Fingerprints™ if they can be determined to be clearly false based on credible, transparent and accountable evidence.
• Documentation: If a narrative is determined to be false, the analyst captures evidence of its falsehood in the form of a written report describing the false narrative, providing evidence proving that it is false, how it emerged, and where it spread.
• Editorial Review and Fact-Check: This explanation and evidence of its falsehood is reviewed by layers of editors, including by NewsGuard’s co-CEOs.

Misinformation Fingerprinting

Once a narrative is confirmed to be false based on credible countervailing evidence, NewsGuard’s analysts use open-source tools and techniques to create a detailed data entry for the narrative containing data seeds designed to enable licensees with AI tools to identify and track instances of the narrative. The fingerprinting process involves capturing dozens of data points about the narrative, such as:

• URLs for examples of content conveying the false narrative from websites, social networks, content platforms, and the open web
• Text and content snippets from examples of the false narrative being conveyed on social networks, content platforms, and the open web
• Related search terms, keywords, and hashtags
• Data about the false narrative’s location and where it has spread
• Data about the false narrative’s origin and start date
• An assessment of the risk of harm for the narrative

False Narrative Monitoring

Once a narrative is fingerprinted, NewsGuard’s analysts monitor for changes in the narrative, variations of the narrative, and new instances of the narrative being published by prominent sources. These data are added to the Misinformation Fingerprint™ for the narrative as they are detected.

At the same time, NewsGuard uses AI-powered monitoring tools to alert analysts when an instance of a false narrative is published by a new, unrated source so that the source can be evaluated for a possible rating.