Does the AI regulation hamper or promote innovation?
Just last month it was reported that Meta will be restricting access to some AI models in the European Union. The same tendency is slowly arising also among other US tech-giants, for example, Apple just recently announced that certain AI solutions will no longer be provided to EU clients. Previous studies also discuss how the AI Regulation is likely to have a negative effect on investments. But what is the actual relationship between the AI regulation and innovation? Does this regulation promote or hamper innovation? The topic will be introduced by Ellex in Estonia experts Hegle Pärna and Merlin Liis-Toomela.
How the AI regulation affects innovation
Many of us probably understand by now, without giving it much thought, that AI systems have come to stay. AI is already being used in many sectors. Various AI systems allow faster production processes in big companies but also change our everyday lives. Everything happening with AI systems has brought up discussions on whether and how AI matters should be regulated in the EU.
Some start-ups and also bigger companies see the AI regulation and the obligations it imposes on AI developers and users as a huge obstacle for innovation. This is mainly because in practice it remains unclear whether and how the model developers are even able to comply with certain obligations they are subject to in relation to AI training. On the other hand, the AI regulation could be said to create opportunities for the safe development of models, thereby making it possible to introduce novel technology. The aim of the regulation is to promote innovation while still protecting fundamental rights. For this, the regulation establishes several measures, the regulation imposes obligations on competent supervisory authorities and member states, for example, the obligation to establish regulatory sandboxes and provide trainings which should simplify things for small and medium sized companies.
Regulatory sandboxes
The AI regulation sets out several solutions that promote innovation so that, on the one hand, it ensures sufficient competitiveness among players of the field but, on the other hand, the requirements under the regulation are complied with. For this, the regulation provides several measures, primarily regulatory sandboxes, which the member states are obliged to establish on a national level. Simply put, a regulatory sandbox is a measure that helps companies test new and novel solutions under regulatory supervision. An example of regulatory sandboxes currently used in Estonia is the Financial Supervision Authority’s Innovation Hub test environment.
It is often thought that regulatory sandboxes are some mythical magic objects that solve most troubles concerning the development of innovative technologies and new regulations. On the one hand, regulatory sandboxes do facilitate the development of new systems by offering several possibilities to create new solutions. Participating in a regulatory sandbox while testing solutions enables to avoid the somewhat strict requirements under the legal acts. Their aim is also to create an environment that facilitates competitiveness while building a bridge between the private and public sector.
Smaller companies, such as start-ups, will have simplified access to such regulatory measures. However, in order to join the regulatory sandbox, the company first needs to provide a document that describes the testing of the novel AI solutions (objectives of the tests, methods, supervision, conduct etc.).
Although the competent authority of each member state needs to ensure that the companies participating in the regulatory sandboxes comply with the provided guidelines, liability for any damage caused lies with the participants. However, the participants of regulatory sandboxes will be spared administrative fines regardless of any damage caused, provided that they comply with the sandbox plan and guidelines in good faith.
What is Estonia’s role?
According to data from various innovation institutions, there were less than 20 regulatory sandboxes in all of EEA, one of these being the above-mentioned test environment of the Financial Supervision Authority. So, the use of this measure is not very widespread in the EEA. The AI regulation gives Estonia the chance to take a step forward in the world of regulatory sandboxes. Namely, by August 2026, each member state needs to establish at least one regulatory sandbox for entrepreneurs to be able to participate if they wish and meet the above requirements. The Swedish data protection authority has already launched a regulatory sandbox pilot project together with other state authorities.
The AI regulation also creates opportunities for cooperation between member states because the member states can establish regulatory sandboxes alone or in cooperation with other member states. This allows smaller member states, like Estonia, to establish quite costly regulatory sandboxes so that not only one state carries the burden. So, in addition to the security and protection of subjective personal rights ensured by the AI regulation, the regulation also gives Estonia the opportunity to partake in the implementation of regulatory sandboxes which is more common in other parts of the world.
Article was published on 03.09.2024 in Ärileht.
