AI Transparency & Regulation

The EU Commission is developing a regulatory framework to promote the EU as a global leader in developing secure, trustworthy, and ethical AI principles. The goal set up the foundations for a European AI strategy while ensuring a well-functioning internal market for AI systems, where the benefits and risks of any AI system are adequately addressed at the European level. 

The so-called AI Act was published as a proposal in April 2021, and it is heavily based on the fundamental EU values and fundamental rights and aims to encourage businesses to develop AI solutions and give users the confidence to adopt them.

Biometric algorithmic systems have depended on artificial intelligence technology for several years now. Although the AI Act is still under development, we would like to address how biometric systems fit in light of the current AI Act draft. All companies commercializing biometric-based solutions in the EU will require to be fully compliant with both the GDRP and AI Act.

A Risk-based approach

The EU Commission has tried to devise a gold standard approach to AI systems analysis by choosing a risk-based approach. The prohibitions and regulations defined by the Draft AI Act are linked to the risks believed to be caused by the respective AI system. 

Let's analyze each level of risk in detail and figure out which applications are deemed to be considered high-risk AI systems.

Unacceptable and high-risk classification can be attributed to solutions in which users' biometric data is used without any user consent and that can compromise their privacy. One example of such systems is real-time biometric identification used for law enforcement in public spaces. Such systems can still be used in case of possible terrorism attempts or any crime scene analysis.

AI systems with specific obligations comprise those systems in which the user is scanned anonymously but without consent. Some use cases included in this risk level are face emotion analysis and biometric categorization, such as gender or ethnicity estimation. 

No risk or minimal risk includes all systems where users give explicit consent or use the system of free will. Such systems include biometric authentication/verification and closed-set identification.

Biometric authentication/verification is currently used in automated hotel check-in, authentication for login, or multi-factor authentication using their mobile devices. 

Youverse solutions are fully developed in the European Union and they are future-proof by ensuring GDPR and AI Act compliance. 

Our solutions support both on-cloud and on-prem integration into your server. We do not store any user pictures, and after the image is manipulated to convert to the biometric template they are immediately discarded. The only entity that can access the created biometric templates is the user that generates them. 

If you're a technical leader or developer, to start integrating and testing, just register and subscribe to a free plan to start testing. Any specific tool you would like to integrate face authentication with? We're integrated with Auth0, and Okta and are continuously expanding this list. Feel free to get help on our discord channel or get in touch directly with our support team.

If you're a business leader or decision-maker and want to know more about how our solution fits in the scope of the AI future in the EU, send us a message or feel free to book a meeting with our team of specialists, who will be happy to help.

About Youverse

Youverse simplifies all customer interactions with private, convenient, and secure face authentication on any device. Ranking Top 5 for accuracy in live face matching in the combined business regions of Europe and the Americas, as independently certified by the international benchmark, Youverse enables a fully handsfree experience, bringing unprecedented levels of convenience and accuracy when adopting face authentication in daily routines.

For more information, visit the company website or its LinkedIn.