Document Trust
On this page you’ll find detailed information about our Document Trust service.
What is Document Trust ?
Yousign enables customers to analyze all types of documents for signs of fraudulent tampering.
To detect tampering we leverage more than 600 fraud detectors that span across three major detection categories :
- Metadata analysis (modification history, suspicious producers, date consistency etc …)
- Structural and visual analysis (fonts analysis, incremental updates etc ..)
- Trust marker analysis (signature, specimen detectors etc ..)
- Advanced AI detection analysis.
To help companies and compliance teams act on this analysis we provide a : Risk Label based on our internal scoring engine.
When using Document Trust, the risk labels are :
| Risk label | Action required |
|---|---|
| High Risk | Reject the document |
| Medium Risk | Manually review the document to assess fraudulent intent. |
| Low Risk | Manually review or accept the document |
| Trusted. | Accept the document |
Each call to Document Trust is also accompanied by a list of indicators that are here to explain and detail the risk label.
These indicators can be used to help your compliance team understand the verdict and take the appropriate decision.
Template based analysis
In addition to the above detectors, Document Trust offers an additional control layer by matching documents to their known structural and visual templates in order to improve the accuracy of fraud detection.
This control layer enables us to return the Trusted risk label on a multitude of common documents.
As we are continuously seeking to improve our document template coverage, please contact our sales team to learn more about the process.
Data extraction on Document Trust
In addition to providing a fraud risk label, Document Trust can provide :
- Extraction
- Document type matching
- Data consistency checks
Yousign uses OCR (Optical Character Recognition) and LLMs (Large Language Models) to turn static documents into structured data and help you automate internal processes in addition to verifying document authenticity.
Document type matching allows you to check that an uploaded document matches an expected document type.
Data consistency checks enable you to verify that information such as : first_name, last_name or other data points are actually present in the document.
Currently these capabilities are available on the following document type list. If you don’t see your document in the list, please contact our sales team or you dedicated TSM to add a new document type.
| Document type | Country of origin |
|---|---|
| social_security (Card or Certificate) | France |
| company_certificate (KBIS) | France |
| business_registration_certificate (Avis de situation) | France |
| driver_licence (Driver's license) | France |
| rne_certificate (Attestation RNE) | France |
| vehicle_registration_certificate (Carte Grise, Libretto Auto) | France, Italy |
| auto_insurance_claims_history (Relevé d'informations) | France |
| temporary_vehicle_registration_certificate (Permesso di circolazione) | Italy |
| tax_notice | France, Italy, Germany |
| payslip | France, Italy |
| invoice | France, Italy, Germany |
| proof_of_address (Certificato Di Residenza) | Italy |
What document formats do we currently accept ?
While we currently support both native pdfs and images (png,jpeg) we strongly advise you to choose native pdfs whenever possible as they keep a stronger digital fingerprint and make it easier to detect tampering attempts.
When pdf alternatives are not available, we still accept the document and perform our fraud detection layer based on image based detectors, but some tampering traces can be permanently lost during the format conversion.
File limitations and guidelines
| Limit type | Limit value | Applicability |
|---|---|---|
| Max file size | 25 MB | All documents |
| Maximum number of pages | 10 pages (check billing unit section for billing informaiton) | All document |
Do not upscale or enhance
Never upscale a low-resolution image, adding extra pixels only adds to processing time without an increase in accuracy.
It is best to avoid very low-resolution images, if possible.
Keep the aspect ratio
Never change the original aspect ratio. Doing so will create distortions and degrade the performance of the OCR.