TruthLens tools
A source can be accurate and still be worth scrutinising, and a source can be careful and still be wrong. This checklist walks through the signals that tell you whether a publisher is accountable, and records which ones you could actually verify.
Paste the URL of an article, then work through the checklist below to see how much evidence you can find for the publisher behind it.
Nothing you type leaves this browser tab. 300 character limit.
Optional, but it makes the report easier to read later.
The author is identifiable
A real person or organisation is credited, and you can find out who they are.
Published and updated dates are shown
You can tell how old the material is and whether it has been revised since.
Claims link to their sources
Numbers, quotations and study results point back to something you could read.
The method is described
How the information was gathered, measured or selected is explained.
The original source is reachable
The underlying report, dataset, paper or recording is available to read.
Independent sources agree
More than one outlet, and outlets that are not simply repeating each other, carry it.
A corrections policy exists
There is a stated way to report an error and a record of past corrections.
Funding and conflicts are disclosed
Who pays for the work, and any interests that could shape it, are visible.
A contact or corrections address exists
Someone answers mail. This is a small signal but a real one.
This is not a credibility rating. It scores how much evidence you were able to gather, not whether the publisher is trustworthy. HTTPS does not prove credibility, a valid domain does not prove its publisher is honest, and a missing byline does not prove misinformation. A high score is not a claim that an article is true.
Every question here has a weight. Signals that are hard to fake carry more than signals that are cosmetic: an identifiable author and reachable citations score 3, a corrections policy scores 1. The percentage is the weight you found divided by the weight of the signals you chose to answer — so leaving everything unanswered gives 0% rather than a judgement.
There is no database behind this tool. It does not look up WHOIS records, domain age, traffic figures, fact-checker verdicts or blocklists. Those checks are either unavailable without a paid API or produce results that look more authoritative than they are. Reading the source yourself produces better evidence than any score.
A national news site: named reporter, publication and update dates, links to the study it quotes, a corrections page, disclosed funding, and two independent outlets carrying the same finding. Method is not described. That scores high on evidence completeness and tells you nothing about whether the specific claim is correct — which is what the SIFT checker is for.
Everything you type stays in this browser tab. There is no account, no server, and no network request — you can confirm that by opening your browser's network panel. Nothing is written to disk unless you choose to copy or download the report.
No. It only covers accountability signals — byline, dates, citations, method, funding. For the structure of a suspicious link, use the URL safety checklist instead. Neither tool performs a reputation lookup.
The percentage only counts signals you answered. If you marked eight of nine as absent, that is genuinely low evidence completeness. If you only answered two, the score reflects just those two.
No. The score measures evidence available to a reader, not the truth of any claim. Accountability is a precondition for trust, not a guarantee of accuracy.
It is a judgement, and it is documented in the scoring rules above rather than hidden. Harder-to-fake signals such as a reachable original source carry more weight than cosmetic ones such as a corrections link.
No, deliberately. Those lookups need a paid API or a public registry that rate-limits aggressively, and a domain age tells you very little about whether today's specific claim is accurate.
Only with care. Different readers find different signals, so scores are not directly comparable across assessors. Use it to structure your own checking, not as a ranking.