Developing Synergies towards an urban-fair AI ACT
On 16 December 2021, the Living-in.EU Legal and Ethics subgroup had the opportunity to discuss the implications and effects of the Artificial Intelligence Act proposal on European cities. The Commission, cities and legal experts shared experiences and views on a legislative proposal that will be essential to ensure citizens' rights to privacy and data transparency, as well as the role of cities, governments and businesses in the implementation and management of AI systems.
The Artificial Intelligence Act (AI Act) - proposed in April 2021 and under consultation until August 2021 — aims to create a common regulatory and legal framework for the 27 countries in the European Union. For this purpose, the framework designates 5 risk levels to AI, and creates a searchable public database of high-risk AI, which will be more heavily regulated, to be maintained by the European Commission. With a view to defining which systems fall under the high-risk category, the proposal will appoint a European AI Board composed of experts from different fields. Cities and their citizens are one of the main beneficiaries of artificial intelligence, but they are also greatly affected by risks that may infringe on their rights, which is precisely what this proposal seeks to ensure. Therefore, for Europe to become a leader in reliable and beneficial AI for its users, it is key that a dialogue with local authorities is maintained and that they develop policies takING advantage of this valuable technology.
Yordanka Ivanova, Legal and policy officer at the EU Commission CNECT A2, clarifies the risk-based approach defined in the proposal. In it, she shows how the range of regulation goes from those systems that do not require any regulation, to those that must comply with some specifications on their transparency or those that will be banned for violating citizens' rights, as is the case of social scoring by public authorities. That is, for AI to be ethical and trustworthy, systems that fall into risk categories must have all existing principles in place to ensure adequate safeguards from design and development. In the case of public authorities, the Commission highlighted several specific requirements to have adequate procedures in place and to ensure that the system is built on good quality data. These include the documentation for the system itself and the guarantee of being traceable, the degree of transparency of information on the functioning of the system to both operators and users, as well as the quality management system and demonstrating compliance with the new requirements.
In the case of cities, the cities of Stockholm, Eurocities and OASC presented local views on the AI Act.
Federica Bordelot, policy advisor on digital policy at Eurocities, points out that there is still a lack of precision on the processing of personal data by AI systems and their actual impact on local governments. While the approach put forward by the Commission is supported, deeper demarcation in each category and more defined and public sector-focused legal requirements remain a concern for cities. In terms of the EU's digital governance, cities demand to be considered for the European AI Board, as well as for more clarity on the responsibilities of the European AI Board so as not to overlap with the European Data Protection Board.
Michael Mulquin, from Open and Agile smart cities (OASC), joins this concern for a fair and transparent application of algorithms able of improving urban life. By doing so, he adds that despite the obligation to comply with these regulations, local authorities still lack the resources and expertise to implement these measures.
Luca Bolognini, President of the Italian Institute for Privacy and Data Valorisation, in his expert capacity, highlighted the complementary nature of this proposal to a European digital framework under construction with other regulations and ethical standards. Outlining the multidisciplinary character of the IA Act, this expert stresses the need to develop teams with competent knowledge for its implementation by local authorities in favour of the intended EU shared community approach.
All in all, this meeting shed light on the debate of creating common ground between European legislation and the needs of cities. These planned meetings will also result in a series of materials, which will be published as they take place. In this case in the field of artificial intelligence. Self-learning algorithms, smart data, sensitive data, privacy, transparency and high-risk uses of AI constitute an ecosystem that both offers and challenges local authorities, which require a European governance system to facilitate their transition.