To the content

thenti - Document Authentication

Solution
Current phase
Ends on:
  1. Call and submissions completed on 12.11.2025
  2. Queries completed on 17.11.2025
  3. Jury evaluation completed on 25.11.2025
  4. Winners announced from 26.11.2025
Announcement of the winners:
26.11.2025

thenti - Document Authentication

thenti offers a security technology, originally developed at ETH Zürich, for authenticating documents in both the physical world (paper documents) and the digital world (PDFs and other formats). The technology can detect minute changes to documents, such as turning an "8" into a "3" or a period into a comma, and thereby uncover subtle document alterations and forgeries. The technology is secure and robust, and it allows users to detect forgeries by scanning documents with standard, commodity smartphones in both controlled environments (high-quality document scans) and uncontrolled environments (poor lighting; folded, creased or soiled documents; etc.).
thenti’s technology works as follows. A document issuer with a document D uses thenti’s server software to issue an authenticated version of D that contains a cryptographically protected QR code. A copy of D is stored, encrypted, on a secured server. The authenticated version of D is then distributed either electronically or physically (printed and sent by postal mail). A document recipient who receives a supposedly authentic document, for example by post, can use the thenti app, running on a commodity smartphone, to scan the document received, which has the thenti QR code superimposed. If the QR code indicates the document should be D, a digital copy of D is pulled (encrypted) from the server to the smartphone and decrypted using a key embedded into the QR code. Then, an AI model running on the smartphone precisely compares the scanned document with the original D. Any changes made to D (forgeries!) are highlighted to the user. For multi-page documents, the visual report is accompanied by a detailed summary describing the number and characteristics of the forgeries found in the document. The user is also notified if the QR code was forged. Because all computation occurs on the smartphone, the privacy of scanned documents is ensured. Moreover, our end-to-end solution has been extensively evaluated by security experts to verify that the workflow meets the highest security standards.
In sum, thenti’s technology ensures document authenticity not by heuristic checks but by precise visual comparison with the authentic original. This comparison can be performed accurately throughout the document’s life cycle (e.g., a physical copy of the document, a picture of a document, a scanned document, photocopies, etc.). The technology is well tested and proven in the field. It has been used in production for almost two years to protect critical documents, such as debt register entries (Deutsch: Betreibungsregisterauszug) issued by the city of Zürich. Over 320,000 city documents have been protected this way, and no forgeries of thenti-authenticated documents have been reported. Finally, the thenti platform is multilingual and has already been deployed in Switzerland, reporting forgeries to users in English, German, French, and Italian.

Added value

The fundamental innovation distinguishing thenti from conventional fraud detection tools lies in our authentication approach. Traditional solutions rely on heuristic analysis, like examining documents for suspicious characteristics, inconsistent formatting, or improbable content. These methods generate false positives and, more critically, fail to detect forgeries that preserve plausible formatting. thenti removes this uncertainty by performing a precise visual comparison of the presented document with the authenticated original, detecting any changes regardless of how plausible the forgery appears. This novel approach provides certainty, not heuristic guesswork: either the document matches the original precisely or modifications are identified and highlighted. The key advantages of this technology include:
Robustness throughout the document lifecycle: Our machine learning models enable accurate authentication regardless of document condition, including folded or creased pages, poor illumination or shadows, etc.
Consistent accuracy across different formats: High forgery detection accuracy, even when a document has undergone multiple transitions between the digital and physical worlds.
Privacy-preserving architecture: All document analysis occurs exclusively on the verifier's device. Scanned documents never leave the device.
Human-understandable differences: Users are presented with intuitive visual reports that highlights alterations directly on the document image.

Files & Info

Queries & Comments

Please log in if you would like to leave feedback.

Solutions already submitted

Strong partners stand behind the Austrian Competence Centre for Innovation Procurement

An initiative of:

In cooperation with:

Cookies

We use cookies to make sure we give you the best experience on our website.

Find out more under "Collection of access data, log files and cookies" in our Privacy Policy.

To the main navigation