AI Act: The EU agrees on rules to support the uptake of trustworthy AI
-
On the 8th of December, the EU institutions reached a provisional agreement on the final form of the AI Act
-
DIGITAL SME welcomes the risk-based approach that is promoted through precise, limited and internationally recognised definitions and thresholds.
-
Very large foundation models meeting criteria that exclude SMEs will also be regulated, thus enabling downstream innovation deployment by SMEs and start-ups
-
In order to enable rather than hinder innovation, practical rules must be developed quickly
In 2021, the European Commission published its proposal for the world’s first regulation of AI. The aim of the AI Act is to safeguard the safety and fundamental rights of users and to boost the AI market by boosting consumer trust in AI applications.
In the past two years, DIGITAL SME has submitted numerous proposals to EU decision-makers for the AI Act to strengthen SME-led innovation, such as an Open Letter sent to Members of the European Parliament, cosigned by over 60 SME associations and innovators from across Europe.
The provisional agreement reflects many of the requests made by DIGITAL SME on behalf of its members. The AI Act contains a precise, limited and internationally recognised definition for what constitutes an AI system as well as for what qualifies as a “high-risk” AI system. However, practical rules must now be derived from this in order to create legal certainty for IT companies.
Furthermore, the AI Act will put in place an ecosystem of support for AI innovation, to promote the deployment of SME-led applications in the market. Regulatory sandboxes will be created to allow innovative SMEs to experiment with products and test their application within a framework of legal certainty. Moreover, European Digital Innovation Hubs and national authorities are tasked to provide technical guidance so that SME AI innovators can become compliant with the Regulation.
The unequal relation between Big Tech AI companies and SME innovators is also being regulated. Digital SMEs will benefit from model contractual clauses can use in negotiations with larger companies to balance out their economic dominance. Very large foundation models that have been trained with computing power above 10^25 FLOPs will be strictly regulated. Such threshold., which likely includes only three US-based companies right now, will allow SME downstream deployers to enjoy legal certainty when building on top of these very large foundation models.
“We welcome the regulation of applications on the basis of risk. We also welcome the regulation of foundation models that dominate the market, allowing digital SMEs to innovate and compete on a more fair basis. But we must not forget that we are introducing a large number of new regulations overall. It depends very much on balanced and quickly available, directly applicable rules for implementation in practice whether we do not end up preventing innovation in Europe again as a result of this AI Act.” says Dr. Oliver Grün, President of the European DIGITAL SME Alliance.
The provisional agreement is a good first step but there remains a lot of work to be done. In the standardisation process as well as during enforcement, the voice of SMEs must be present if Europe wishes for AI innovation to flourish. DIGITAL SME remains committed to monitoring these processes and ensure that Europe meets its Digital Decade targets, especially when it comes to SME AI innovation.