DOCUMENT AUTHENTICITY / RESEARCH & TOOLS
Look beyond
the document.
Understand fake IDs and forged documents. Learn what verification can prove. Test your software with data that can’t be mistaken for identification.
Independent education. Fictional data. No identity documents for sale.
One check. One limited answer.
Are the fields consistent?
Is the document genuine?
Does it belong to this person?
THE READING ROOM
Make better verification decisions.
Fake IDs: the problem behind the label
Counterfeit, altered, borrowed and stolen documents require different responses.
Read the field guide 02 / DETECTIONA suspicious document. What next?
A practical triage checklist, with clear limits on visual checks and escalation.
Work through the checklist 03 / SECURITY FEATURESRead the evidence, not the appearance
Understand layers of document security and what an ordinary image leaves out.
Explore security featuresTHE TEST BENCH
A test input.
An expected result.
A useful comparison.
Build a fictional scenario pack, check a JSON or CSV batch, and compare the output from your application. Seven free browser tools for repeatable QA, including an OCR image pack with ground truth and regression scoring.
Local processing · No account · No uploads{
"classification": "FICTIONAL QA — NOT VALID",
"case": "missing-name",
"expected": {
"outcome": "fail",
"issueFields": [
"fullName"
]
},
"validForIdentification": false
}INDEPENDENT RESEARCH
Check the claim before choosing a tool.
OnlyFake: what the documents establish
Read the February 2026 official record, current-domain uncertainty and bounded QA options.
Read the research note OFFICIAL RECORDVerifTools: a dated evidence trail
Separate the official 2025 reports from unverified claims about present domains.
Inspect the timeline TOOL SELECTIONA fake generator for which task?
Choose marked records, abstract OCR sheets or arithmetic cases by the behavior you need to test.
Choose a test assetA BETTER STARTING POINT
Useful evidence.
Honest limitations.
Sources you can inspect
Guides link to standards and official resources. A claim should be traceable, and uncertainty should stay visible.
Checks you can understand
Tools explain their calculations. A passing result is never presented as document authentication.
Research you can reproduce
Our first dataset pairs checksum collisions with explicit test cases, so you can inspect the result yourself.