What this AI fact checker does
It separates a response into checkable pressure points: exact numbers, dates, named people or studies, current rules, causal explanations, recommendations and promises. It helps you choose the next source; it does not invent a truth score.
Why confident wording is not evidence
Words such as proven, always, guaranteed, official, latest and safe for everyone raise the cost of being wrong. They are useful search targets, not proof that the answer is false.
Check the source, not only the citation
Open the original page, find the exact passage and compare its date, population, location and wording with the claim. A real link can still be irrelevant, outdated or too narrow for the conclusion.
Use it for current information
Prioritize prices, laws, public office holders, product availability, software versions, schedules and medical or financial claims. These can change after a model learned its training data.
Keep unknown separate from false
If a source is missing, inaccessible or ambiguous, mark the claim unclear. Honest uncertainty is more useful than forcing a binary answer from incomplete evidence.
Private by default
The quick review runs in this browser. The original answer is not uploaded or stored; only an optional summary receipt can be shared. For consequential work, request a bounded evidence review after the free pass.
Frequently asked questions
Hogar Mas · hogarmas.net
Can an AI fact checker prove that a claim is true?
No. It identifies claims that deserve verification and organizes the review. You still need a suitable primary or authoritative source.
Can I fact-check a ChatGPT, Claude or Gemini answer privately?
Yes. The local check runs in the browser and does not upload the pasted answer. Text only leaves the page when you explicitly open an external search or share a summary.
What should I verify first in an AI answer?
Start with current rules, exact figures, dates, named studies, safety promises, citations and claims that would change an important decision.
Does this identify which AI model wrote the text?
No. It reviews the claims rather than guessing the author or model.