“Catalonia is ready to successfully face the challenge of developing and deploying artificial intelligence in an ethical, responsible and reliable manner which, in turn, is fully compatible with the promotion of innovation in this field”. So stated the Secretary for Digital Policies of the Government of Catalonia, Gina Tost i Faus, on the occasion of the publication of the new European AI Act in the Official Journal of the European Union (OJEU).

It is an innovative law that promotes more responsible and sustainable AI and aims to harmonise AI standards across the European Union. It is considered the first standard of its kind in the world and therefore a global standard for AI regulation.

The new law, which the Council of the European Union passed last May, will enter into force on 1 August. From then on, the different companies and organisations that develop, generate or use AI systems will have to adapt to the regulation. This means that 6 months after its entry into force, prohibitions will applyon “unacceptable risk” uses of AI, such as those involving AI systems that manipulate people’s decisions or exploit their vulnerabilities, or systems that predict a person’s risk of committing a crime, among others.

As Secretary Tost said, “the Government of Catalonia has been working for years to position the country as a leader in the use and development of AI technologies in the public and private sectors and, at the same time, to build trust in these technologies and prevent unintended consequences that could harm individuals or communities”.

And it does so through Catalonia’s Artificial Intelligence Strategy, Catalonia.AI, which has among its fundamental pillars the Artificial Intelligence Ethics Observatory of Catalonia (OEIAC). This is a body that studies the ethical, social and legal consequences and the risks and opportunities of the implementation of AI in daily life in Catalonia, from a fully cross-cutting perspective.

Tost said that “innovation and regulation are not polar opposites, but a compatible framework for charting a safe course”. In this regard, she stated that “the defence of people’s digital rights is a priority for the Catalan Government, which is why we promote projects and provide tools for the ecosystem, such as the new PIO Model for the assessment of risks associated with AI or the resources generated by the Aina Project, which are secure, transparent and fully aligned with the AI Act”.

The PIO Model, a tool for assessing risks associated with AI as a public service

The new PIO (Principles, Indicators and Observables) Model developed by the OEIAC is a pioneeringtool both in terms of its conception and its public service and open access nature. It contains ethical recommendations but also legal obligations, promoting compliance with current AI rules and regulations through a thorough verification process. It is now openly available to all.

It allows appropriate or inappropriate actions to be identified and provides tools for the quadruple helix (administration, academia, business and citizens) to encourage ethical and responsible uses of AI data and systems. In addition, along with verification, it incorporates other resources, such as ethical recommendation clauses for the purchase and procurement of AI systems, which serve as a guideline for public and private organisations to incorporate in their procurement processes.

Catalonia.AI, a strategy aligned with the principles and obligations of the AI Act

Based on the conviction that the Administration must be exemplary in the deployment of AI, all initiatives and projects developed within the framework of the Catalonia.AI strategy are aligned with the ethical principles and legal obligations set out in the AI Act.

An example of this is the Aina Project, promoted by the Generalitat de Catalunya in collaboration with the Barcelona Supercomputing Center (BSC). It aims to generate corpora and computer models of the Catalan language so that companies creating AI-based applications, such as voice assistants, Internet search engines, automatic translators and correctors or chatbots, can easily do so in Catalan. The resources worked with the Aina Project are high quality, secure and transparent in terms of data and training processes; can be assessed and explained, and are open, permissively licensed, and available to all through the Aina Kit.