
NLPAICS 2026
Bloc-Conditional Event States: Measuring Cross-Coverage Divergence for Threat-Intelligence Analysis
Maryam Fooladi, Federico Bottino
Paste any article, post or publisher. Newjee measures what it claims, how it frames it, and how differently the rest of the world is covering the same story.
Sign in to run a Reality Card. See how it works
Breaking news · A leader addresses the crowd
Illustrative footage, AI-generated. Not a real event.
Newjee's methods are published. Three of our papers were accepted in 2026, at NLPAICS and at two LREC workshops, and are on the ACL Anthology and arXiv.

NLPAICS 2026
Maryam Fooladi, Federico Bottino

PoliticalNLP 2026 · LREC 2026
Maryam Fooladi, Federico Bottino

InDor Workshop · LREC 2026
Maryam Fooladi, Federico Bottino
You shouldn't have to take a verdict on trust. Newjee shows where each number comes from, so you can check it yourself.
The AI labels each claim as verified, unverified or contradicted. The score is calculated from those labels, so it always matches the claims you see. When fewer than three claims are settled, or the unverified ones outweigh them, Newjee shows no score. A topic search gets no score, since a person or a meme can’t be graded; when the topic is itself a claim, such as a theory or a rumour, you get a verdict on that claim.
Each step of a story’s spread is marked Observed if a page from our web search backs it, or Inferred by the model if none does. Newjee reads current coverage, not an AI’s old training data.
Newjee doesn’t give outlets a trust score it hasn’t measured, and doesn’t cite ratings or awards that aren’t in the coverage. Wire copies count as one report, not several.
Paste an article or type a topic in any language and the analysis comes back in English. Outlet names stay exactly as they are published.
What Newjee does, how it works, and how to get involved.
Question we didn't cover? We answer them ourselves.