“Space. The final frontier.” The year is 2265. USS Enterprise has been deployed to explore far-away galaxies. On board of the starship is a 400-member crew, plus an on-board computer without which practically nothing works. Based on its knowledge of facts, it issues recommendations for actions it has generated by performing probability calculations. What was a mere vision in the “Star Trek” science fiction television series from 1966 has increasingly become a reality today: Human thinking and learning serves as a model for and provides impetus to an Artificial Intelligence with the ability to make sensible choices. In a concept study, Schreiner ProSecure and Schreiner Digital Solutions investigated how AI may be used for smartphone-based identification of security features in labels.

AI has become an integral component of everyday life, for instance in intelligent personal assistants and navigation systems. In industrial process automation, it has been gaining traction, too. In smartphone cameras, it’s used for image recognition, which the study was focused on as well: How can security features be identified using AI-based image recognition technology? How can hidden misprints be detected which are invisible to the human eye?

For the feasibility study, the investigators used ShiftSecure, an overt security feature produced using inks from banknote printing that generates a light-dark or positive-negative effect depending on the incidence of light and angle of vision. Commercially available smartphones and a modular software concept that also checks further security features were used for authentication. The objective of the study was to find out how precise AI-based authentication can be.

Für die Machbarkeitsstudie wurde ShiftSecure verwendet – ein offenes Sicherheitsmerkmal, hergestellt mit Farben aus dem Banknoten­druck, das je nach Lichteinfall und Betrach­tungswinkel einen Hell-Dunkel- bzw. Posi­tiv-Negativ-Effekt erzeugt. Zur Authentifizierung wurden handelsübliche Smartphones und ein modulares Softwarekonzept genutzt, das auch weitere Sicherheitsmerkmale prüft. Ziel der Studie war es, herauszufinden, wie exakt eine Originalitätsprüfung mittels KI sein kann.

Compared to a conventional verification solution, AI-based authentication offers a wide variety of benefits, especially in terms of its compelling robustness: correct results are achieved even in poor lighting conditions or when using lower-end smartphones. There’s no need for training users because the security feature can be recognized automatically. A special algorithm is used for analysis. It automatically checks the features that otherwise require tools for testing. The system provides automatic feedback to the brand owner whenever it identifies a fake.

The outcome of the study: ShiftSecure is suitable for AI-based detection and authentication with 99-percent accuracy, which may be raised in combination with further security features. While still representing a concept study at this juncture, the analysis provides the basis for developing a pioneering solution for customers in the Mobility and Healthcare industries for protecting the authenticity of their products.

Das Ergebnis der Studie: ShiftSecure kann KI-basiert erkannt und mit einer 99-prozentigen Genauigkeit authentifiziert werden. Diese Quote kann in Kombination mit weiteren Sicherheits­merkmalen erhöht werden. Noch handelt es sich bei der Auswertung um eine Konzeptstu­die. Doch sie bietet die Grundlage, um für Kun­den aus der Mobilty- und Healthcare-Industrie eine zukunftsweisende Lösung zum Originali­tätsschutz ihrer Produkte zu entwickeln.