On 25 January, EIF organised a hybrid debate focused on “The EU Artificial Intelligence Act: the right framework for trustworthy AI and innovation?”, an opportunity to discuss the draft regulation from the commission, the first comprehensive attempt at global level to regulate specific uses of AI systems. The debate was hosted by MEP Brando Benifei and featured the following speakers:
- Yordanka Ivanova, Legal and Policy Officer at the unit Artificial Intelligence Policy Development and Coordination, DG CNECT, European Commission
- Julien Vignon, AI Policy Lead, The French Ministry for the Economy and Finance
- Andrea Renda, Senior Research Fellow and Head of Global Governance, Regulation, Innovation & Digital Economy, CEPS
- Frederico Oliveira Da Silva, Digital Rights Senior Legal Officer, BEUC
- Emilia Tantar, Chair of DIGITAL SME's Focus Group Artificial Intelligence
MEP Benifei opened proceedings by stating that the MEPs aim to strengthen the legal certainty of the text, calling for clearer definitions, in order to give a real message of trust and understanding of these upcoming changes. Moreover, the regulation should be clear also on the balance between respect and defence of fundamental rights. According to the MEP, the Parliament has an important role to play and it is important for citizens, consumers and organisations alike to feel safe with the usage of AI.
Regarding the AI Act and other pieces of legislation, MEP Benifei stated that the regulation will intertwine with all data protection regulations in place, not only GDPR, but also the Digital Services Act and the Digital Markets Act that are now under negotiations.
In Yordanka Ivanova’s opinion, the AI Act is an unique opportunity to make Europe a global leader in trustworthy and human-centred Artificial Intelligence. The principles that the Commission based the regulation on are building a proportionate, balanced framework, that follows a risk-based approach, therefore regulates only where strictly necessary. The regulation should remain flexible and future-proof, as it is dealing with a technology that is developing very fast.
Ms Ivanova reiterated the Commission’s commitment towards supporting innovation and SMEs, access to regulatory sandboxes in order to create the right conditions and tools which ensure all Member States can work together.
Julien Vignon welcomed the proposed risk-based approach and highlighted the opportunity for Europe to tap into the fantastic potential of AI while protecting its citizens’ rights, which requires a legal framework with clear rules that respect innovation capacities that are in sync with our European values.
The French Presidency made some proposals in the direction of more clarity and predictability and, at the same time, the need to think about easy enforceability as much as possible, to not put undue burdens on providers. Mr Vignon highlighted the importance of trust for ensuring high uptake but also preserving and enabling innovation capacities.
The AI Act is a good first step, but still incomplete, according to Andrea Renda. A risk-based approach could be extremely conservative, very constraining on innovation, or it could be well timed, adequately defined and adaptive, for innovators to know exactly how much they should worry about potential impacts depending on the application that they are developing.
Mr Renda considers the process as the most important part of the AI Act, because we should avoid having a too prescriptive piece of legislation that will be obsolete by the time it hits the market. Therefore, a governance system with adequate resources, coordination, and consistency with the other pieces of legislation is needed.
Frederico Oliveira brought to the table the consumer perspective. As the first legal text regulating AI, BEUC welcomes it but insists that the consumer dimension and perspective is still lacking and needs reinforcement. Mr Oliveira agreed there should be a risk-based approach, starting at the basic layer, and that more ambition and clarity is needed about the AI practices that have been banned (like social scoring or emotion recognition). Moreover, consumers should be entitled to the rights to transparency and to redress.
Emilia Tantar, being part of the Digital SME Alliance’s AI focus group, identified some challenges for SMEs in the adoption of AI. These included access to finance data, skilled personnel and having a strong, solid ground, which came in the form of the Artificial Intelligence Act. Ms Tantar considered this legal proposal the solid ground that SMEs needed in order to thrive in AI. Together with the AI task force within the Focus Group on AI, she identified several key points for improvement of the AI Act: 1. A need for more clarity in the definitions, 2. Conformity assessment and compliance costs, 3. Standardisation, 4. Liability all over the value chain, 5. Access to testing environments.