Account & how this works
This app does not say “trust us”. It says: here's the evidence, here's the uncertainty, look for yourself.
Loading…
The free plan includes 5 investigations, 1 deep investigation and 2 screenshot investigations each month. Pro raises the allowance, unlocks deep and screenshot investigations whenever you need them, and keeps your saved cases.
Live search, source retrieval and screenshot reading are connected. Screenshots are read by an AI vision model that transcribes only the visible text — the image is treated as material from an untrusted source and checked against outside evidence, never taken as proof on its own. The completed case files in Explore are clearly labelled demonstration cases, written by hand.
Evidence over verdicts
Simplistic true/false labels are avoided whenever the evidence is more complicated than that. Statuses describe the state of the evidence, not a side.
Sources are ranked, not counted
Primary documents, official data and court or regulatory records come first. Multiple outlets repeating one original report is not independent confirmation.
Posts are evidence of what was said
A social post proves someone said something. It does not, by itself, prove that what they said is true.
Incentives are context, not proof
Benefiting from an event is never treated as evidence of having caused it.
Care with people
Allegations about identifiable people are attributed, not asserted. “Person X was accused by…” rather than “Person X did…” unless the evidence is strong.
No confidence theatre
No invented percentages. Confidence is Low, Moderate or High, and the reason is always stated.
Same framework for everyone
No party, politician, government, ideology or bloc gets a different standard — and strong evidence is called strong even when one side dislikes it.
Cases marked DEVELOPING carry a standing warning: early reporting can change, and some details may not yet be independently confirmed. No conclusion is forced just because one was asked for.