GoldenCard and VehicleOwnershipCertificate document typesVehicleNumber, PassportNumber, and TrafficParticipantNumber to the v3 field types.injectionAttackCheck group and sdkPayloadIntegrityCheck for multipart SDK payloads, validating signed images and request fields and returning Reject when tampering is detected. This is exclusively available for v3.Verify is now aligned with the latest data extraction capabilities in BlinkID.
Starting with this release, Verify uses the epoch versioning scheme:
{EPOCH * 1000 + MAJOR}.MINOR.PATCH
The epoch increments on significant or groundbreaking changes, so the version after 3.22.3 is 4000.0.0 rather than 4.0.0. Subsequent releases follow the same scheme.
We're introducing a new v3 endpoint.
The new endpoint is at /api/v3/verify.
Both the request and response surface has been significantly changed, with numerous fields being renamed. Some important changes include:
verdict in the response.failedChecks array that lists any checks that the document has failed.passesAtOrBelowSensitivity, the highest sensitivity at which they would still pass.imageAssessment.imageQualityCheck result.
The same module is now used by both BlinkID and Verify.pipeline object with a separate status for the extraction and verification stages.portraitForgerySensitivity has been re-tuned to Level5.apiImageTag, workerImageTag, gatewayImageTag, identityImageTag, executionId, consentIdWe now require a consent object.
You now must provide an end user's consent if you want to process their data.
The minimum information we require from you is the duration of the consent, and a unique user ID:
{
"durationDays": 60,
"userId": "634474d4-9577-..." // we recommend using UUIDs
}
For more information on consent, why it is required and how it works, see Get your users' consent.
See the migration guide from v2 to v3 for a detailed breakdown of everything that has changed.
The extraction engine is updated to BlinkID v8001, which adds support for numerous new documents and versions, and improves OCR accuracy.
See the full list of supported documents here.
verificationContext is set to InPerson, since a document presented in person doesn't need a synthetic-image check.vehicleClassInfo, driverLicenseInfo, and sex from barcode to the v2 extraction response for driver's licenses.With this version, we're releasing the Verify web SDK.
Install it:
npm install @microblink/blinkid-verify
yarn add @microblink/blinkid-verify
pnpm add @microblink/blinkid-verify
You can use the Verify web SDK for handling document scanning and recognition in web applications, simplifying the integration with the Verify API.
We're introducing new, session-based analytics to provide a more granular view of the entire verification journey, rather than only verification outcomes. This will help us give our clients insights into user success, friction points, drop-offs, and a complete funnel analysis. Analytics don't collect or store personal data. The system captures only operational and behavioral events to improve performance and usage.
We've adjusted the internal document verification logic to expand the use of our Photocopy Detection machine learning model. This improvement allows photocopy detection to be applied to the back side of ID documents as well, increasing fraud detection coverage while maintaining the same overall verification flow.
The structure of our API response has been refined to improve clarity and readability of the verification results. This is the first iteration of ongoing improvements aimed at making responses easier to understand in production. This isn't a breaking change: only the order of fields has changed.
Verify is now aligned with the latest data extraction capabilities in BlinkID version 7.7, ensuring that any document supported by BlinkID can also be extracted via Verify. This provides broader document coverage.
Security vulnerabilites fixes
]]>generativeAiCheck and generativeAiMatchLevel in the API schema to configure whether the check runs and how strict it is, consistent with other checks.CaptureConditions::Hybrid that allows Accept for cropped images (skipping liveness)contentCheck for British Columbia documents when the first or last name is missing and the barcode cannot be parsed.Hybrid value to the CaptureConditions optionAwaitingOtherSide in case of US passportsimageFront / imageBack / imageBarcode are provided as empty strings, the API will now return 400 Bad Request. Use null when no image is provided.Example request (using null for missing images):
{
"imageFront": {
"url": "image"
},
"imageBack": null,
"imageBarcode": null,
}
Note: Instead of "url" you can also provide a "base64" field:
{
"imageFront": {
"base64": "/9j/4AAQSkZJRgABAQE..." // cropped
}
}
Example of an invalid request:
{
"imageFront": {
"url": "image"
},
"imageBack": {
"url": ""
},
"imageBarcode": {
"url": ""
}
}
FirstSideAdditionalProcessingInfo and SecondSideAdditionalProcessingInfo fields (of type AdditionalProcessingInfo) to the AdditionalInfo object in the V2 APICertificateNumber, CountryCode and NationalInsuranceNumber to BlinkIdMultiSideRecognizerOutput and VIZResult (V1 API and V2 API)BarcodeAnomalyCheck will no longer Fail in MultiSide mode for unsupported documents but result in NotPerformeduseCaseAdded the following fields to the response
useCaseUsedoptionsUsedrecommendedOutcome inside Verification objecttraceIdFullyCroppedDocument value added to the ProcessingStatus fieldDataLogic
documentDiscriminatorLogiccustomerIdNumberLogicinventoryControlNumberLogicDocumentNumberLogic, PersonalIdNumberLogic and DateLogic checksMatchCheck bugs caused by permanent and partial datesBlurCheck from ImageQualityCheckImageQualityInterpretation which allows setting a desired image quality strategy, ranging from high assurance to high conversionreadCheck and contentCheck to BarcodeAnomaly resultImageQualityInterpretation to control how ImageQualityCheck results impact the decision to make an overall fraud verdictoverall and clarityCheck to ImageQualityCheck resultversion object in response, it returns API version and BlinkId Verify core versiondataIntegrityCheck to DataCheck memberimageFormat optional parameter. The default value is PNG
JPG for improved performance in request timeAnomalyCheck from VisualCheck memberreturnAnomalyHeatmap optional setting from requestdocver/multipart to support multipart form data