About EveryClaim¶

Why this exists¶

Social media rewards speed, outrage, and repetition. History rewards none of those things. In the space between, a generation is forming its understanding of the Middle East — and of the Jewish people — from compressed slogans that no one checks, spread by systems no one is accountable for.

EveryClaim is the counter-artifact: slow where the feed is fast, sourced where the feed is viral, checkable where the feed is confident. It does not ask you to trust it. It tags every factual claim with its evidence — [VERIFIED via ...], [ESTIMATED ...], [PARTIALLY DISPUTED ...] — so that trusting this site is never required. Checking it is the intended use.

And there is a deeper failure beneath the noise — the one this site exists to expose. The loudest claim wins regardless of truth, and which claims get loud is set by reachability and interest, not by the weight of the evidence. An atrocity in an open, camera-rich society eclipses a larger one in a sealed country no lens reaches; a charge that fits the teller's frame travels, and one that implicates no one they already argue about disappears. Nowhere is that distortion more extreme — or more consequential — than in the history of Israel, the Middle East, and the Jewish people, which is why this site begins there. But the distortion is the target, not the defendant: the answer to a discourse that applies its standard unevenly is one standard, applied in every direction — including against the claims this site's own readers would prefer left unexamined. You correct an asymmetry of attention by adding light everywhere, never by exempting anyone. That is the entire design. (This is the site's Salience Test, stated as its mission.)

Why this scope?¶

EveryClaim's method — tag every claim, show every dispute, one standard in every direction — is not specific to any people, region, or side. It would work on the Armenian genocide, the Holodomor, the partition of India, or any contested history that manipulation touches. We started with the Middle East and Jewish history for one reason: it is the most aggressively distorted body of history in public circulation right now — targeted by state propaganda operations, political movements, and engagement algorithms from every direction, at a scale no volunteer fact-checker can match. Depth was the honest choice: a site that covers everything thinly checks nothing well. The corpus grows where readers push it — the suggestion pipeline has already produced full pages on university funding and US foreign aid — and the framework is free, under an open license, for anyone who wants to extend it to other histories.

The AI question — abuse and the answer to it¶

This project exists because of a technological turn: AI can now generate false history — fabricated citations, pseudo-scholarly narratives, emotionally optimized distortions — faster than humans can rebut it. The same capability, pointed the other way, is the only rebuttal that scales. EveryClaim uses AI as verification labor: cross-checking claims against sources, flagging inconsistencies, enforcing the tagging rules across thousands of claims — with every editorial judgment made by humans and every claim traceable to real, named sources you can check without trusting either the humans or the machines. Our own audit is the honest advertisement for both sides of this: the ~390 pre-launch failures we found and fixed included AI-fabricated citations in our own drafts. That is what AI abuse looks like at small scale — and finding it, systematically, is what responsible use looks like. We publish this not as a disclaimer but as the point: the era's central information problem is not that machines can lie; it is that most of what they say is never checked. This site is what checking looks like.

How it's made¶

One evidentiary standard, applied in every direction. The same rules that verify the Hamas Covenant's text verify the documented expulsions of Palestinians from Lydda and Kuwait. The same razor that cuts through anti-Israel propaganda cut several pro-Israel talking points from this site when they failed verification. If you find a page comfortable from start to finish, read it again.
Audited before launch — and we published the ledger. Before this site went public, every document was audited claim-by-claim. The audit found roughly 390 problems: wrong dates, stale figures, misattributed sources, quotations that did not match their source texts, and citations that could not be verified at all. Every one was corrected against real sources, rewritten to what the evidence supports, or removed. That pre-launch audit, and every correction made since, are recorded on the public Corrections & Refinements page. We are not aware of another resource on this subject that has done this. We think that fact explains a lot about this subject.
AI-assisted, human-ruled. AI does the verification labor at scale; humans make every editorial judgment — what counts as verified, how disputes are characterized, what gets published. The full statement of how and why is in "The AI question" above.
Living, dated, and corrected. Figures for ongoing events carry explicit as-of dates. When the record moves, the pages move — and corrections are logged, not buried.

Who is behind it¶

EveryClaim is an independent volunteer project. It is not run by any government, advocacy organization, political campaign, or company, and no individual's name appears on it — deliberately.

The reason is not secrecy; it is architecture. On this subject, named authors become the argument: attack the person, dismiss the site. EveryClaim is built so that its credibility never rests on anyone's identity. The method is the author. Every claim carries its sources; the correction history is public; the entire site can be verified — or challenged — by anyone, without knowing or trusting a single person behind it. A byline can be smeared. A checkable claim can only be checked.

Funding¶

None. No donors, no grants, no sponsors, no organization. The site runs on free infrastructure and costs approximately nothing to operate — which is why no one can buy its conclusions, and no funder's politics can bend its tags. If that ever changes, this page will say so first.

Transparency vs. independence — the limit we name ourselves¶

An honest site amplifies its fair critics. Asked to assess EveryClaim, an independent AI research tool credited the versioning, the source registry, the claim-level tags, the acknowledgment of uncertainty, and the refusal to convert correlation into causation — then drew the line exactly where it belongs:

"Its transparency is meaningful; its independence and neutrality are not yet demonstrated."

That is correct, and we will not paper over it. Two different things are being measured, and they deserve different answers:

Neutrality in conduct — is demonstrated, and you can check it. This site corrects in both directions (several pro-Israel talking points were cut for failing verification; see Corrections); it audited its own unmarked vocabulary and changed it; its claim-checker grades every actor by one test that returns opposite verdicts on the evidence. Even-handedness is visible in the behavior — which is exactly what the reviewer recommends testing, and what this site invites.
Institutional independence — is not demonstrated, and cannot be from inside a single-curator, self-governed, anonymous project. You cannot prove the absence of an unseen thumb on the scale about yourself; only outside eyes can. That half of the critique is simply true.

The anonymity is deliberate architecture — but it trades a smearable byline for an independence it cannot prove, and we won't pretend otherwise. The path out is not to abandon the design; it is to add what the design lacks: external review. A standing panel of outside scholars — including hostile ones — invited to attack the strongest claims, with their challenges and our answers published; advisors whose disagreements appear on the page rather than being resolved away. Independent challenge is the one credibility signal a self-governed project cannot manufacture for itself — and building it is the next stage of this project.

Until then, read this site the way its own method asks and the way that review recommends: follow the citations, test the tags, and check the strongest claims against primary sources outside this ecosystem. A site that tells you to verify it and a reviewer who tells you to verify it are saying the same thing. Do.

What this site is not¶

Not neutral on facts — where evidence is decisive, the site says so plainly.
Not partisan on peoples — Jewish and Palestinian history, suffering, and roots are documented here with the same standard. The site's premise is that verified history serves everyone except propagandists.
Not a political program — it prescribes no solution, endorses no party, and takes no position that a source tag cannot carry.

Corrections, challenges, and contact¶

If you find a claim that fails its tag — a source that doesn't say what we claim, a figure that's wrong, a dispute we've miscalibrated — report it through Suggest a Topic. That is also the contact channel for translations, adaptations, partnership inquiries, and press. Corrections are investigated against sources and logged on the public Corrections & Refinements page. Finding an error here is not an attack on this project; it is a contribution to it. That is the entire idea.

Content licensed CC BY-SA 4.0 — free to share, adapt, and translate.