The use of Artificial Intelligence (AI) solutions is vital to global competitiveness, security, and the ability to provide quality customer experiences and relevant services to citizens. The European Commission’s proposed AI Act builds on the concept of trust. Building trust is the fundamental point of any AI regulation with the aim of creating the best possible climate for innovation and investment to shape Europe’s digital future. Unfortunately, the good intentions of the Commission’s proposals will be lost in practice. The current version of the proposal is indeed risky for investments and competitiveness, and will jeopardize innovation, and AI rollout in Europe. Therefore, the proposal needs to be significantly revised.
Technological development benefits from international research and trade. Swedish industry is at the forefront of many markets due to its ability to buy components and AI solutions from outside Europe. Our exporting businesses build their administrative compliance structures in ways that work in all the countries where they operate. Therefore, regulations on the use of technology need to be based on international principles and agreements to promote international trade. The OECD’s AI principles and the Ethics Guidelines for Trustworthy AI are therefore the right starting points. But the EU Commission’s AI Act overlaps with existing frameworks in a way few companies are able to address. On top of that several requirements on data processing do conflict with the GDPR and are therefore unlawful. As it is currently drafted, the AI act will place disproportionate costs and administrative burdens on many businesses, and especially on AI innovative SMEs and start-ups.
There are substantial concerns that the AI Act proposal is far too comprehensive and that it risks undermining European competitiveness and international trade. Significant adjustments to the proposal will be crucial to ensure that it is practically applicable to high-risk AI systems, in a proportionate manner and is consistent with existing legislation.
AI is already regulated in several policy areas, but some aspects of AI solutions challenge existing regulations, for example autonomous systems, self-learning, different actors involved during the life cycle, open and changing systems, and the black box problem. These challenges need to be address by existing regulations to avoid conflict between regulatory systems and to establish predictable, applicable, and consistent laws. The AI Act overlaps with data protection, product safety, discrimination and liability frameworks, and needs to be revised to meet the better regulation principle and ensure compliance.
The AI Act must not be another regulation that creates barriers to trade and reduces entrepreneurs’ opportunities to create new solutions and businesses. We should strive to strengthen the competitiveness of European companies, not undermine it. It is now up to the European Parliament and EU Member States to amend this proposal to support the uptake of AI and the EU’s technological capacity.Artificiell intelligens