The EU has started to regulate AI. The EU's AI regulation comes into force on 1 August.
The EU has started to regulate AI. The EU's AI regulation comes into force on 1 August.
The EU has started to regulate AI. The EU's AI regulation comes into force on 1 August.
On 1 August, an EU regulation comes into force that regulates the use of artificial intelligence systems within the EU. The act, which aims to ensure that Europeans can trust products connected to AI systems, also comes with a number of obligations. For non-compliance, companies will be penalised up to €35 million or 7% of their worldwide turnover.
The regulation will affect all companies wishing to place their AI products on EU markets, as well as European companies using AI-connected products. Any machine-based system that can operate autonomously on some level and can learn from the input it receives, is an AI system and will have to comply with the obligations.
As the risk increases, so do the requirements
"The regulation divides AI systems into several risk categories, with higher risks associated with stricter measures," explains Marek Priesol, Legal Partner at BDO. The highest risk group includes products that are already covered by EU product safety legislation. This includes toys, medical devices, cars and aviation equipment, for example. In addition, the Act defines a number of other specific product areas that will have to be registered for assessment before they can be placed on the market. Products used in the fields of critical infrastructure, education, law, migration, employment and others will not escape this obligation.
"Prohibited practices" within AI systems
In addition, the regulation also contains a list of "prohibited practices" used in AI systems, which 6 months after the entry into force of the law, i.e. from 1 February 2025, can no longer be used within the EU. "Six months from now, your business may no longer be able to use some of the AI systems it uses today, so if you are using AI, we recommend that you assess your options for continued use in relation to this legislation," Priesol explains.
In addition, the Act will not only affect users of systems by potentially restricting its vendor. The regulation also sets out requirements for demonstrating transparency in the use of such technologies as well as the generation of content. Businesses that use AI are “deployers” in this sense and will mostly have to be transparent about their use as well as the content they generate. For non-compliance, companies can be penalised up to €35 million or 7% of their worldwide turnover. Public authorities will also be subject to obligations. However, areas such as military military, defence or matters of national security are exempted from the regulation.
Only practice will tell whether European regulation will enable the healthy and controlled development of artificial intelligence, or whether it will, on the contrary, hinder economic development and put the European market at a disadvantage compared to less or not at all regulated jurisdictions.