On 14 June 2023, the European Parliament adopted a negotiating position on Artificial Intelligence (AI) law. The European Parliament’s vote on the AI law proposal will set the standard for inter-agency negotiations and is an important milestone towards AI regulation within the European Union, as further explained below.
The proposed AI law follows a risk-based approach, banning AI applications that pose unacceptable levels of risk and imposing strict regimes on high-risk use cases. It also sets out obligations to AI providers and adopters of AI systems that are aligned with the level of risk posed by AI.
The European Parliament has expanded its initial list of AI systems with unacceptable levels of risk to include a ban on intrusive and discriminatory uses of AI. This includes real-time remote biometric identification systems in publicly accessible spaces, biometric classification systems using sensitive traits, and more. gender, race, ethnicity, citizenship status, religion and/or political orientation), predictive police systems (based on profiling, location, or past criminal activity), law enforcement, border control, workplaces and educational institutions Create facial recognition databases by scraping facial images from emotion recognition systems or uncovered Internet or closed-circuit television (CCTV) footage.
Providers of “foundational models” (i.e. AI models trained on large amounts of unlabeled data that can be adapted to many applications) assess and mitigate potential risks to health, safety or fundamental rights and The model must be registered in the EU database. Before launching on the EU market. In addition, generative AI systems based on such models are subject to a set of transparency requirements, including the obligation to disclose when content was generated by AI and to safeguard against the generation of deep fake content. You are obligated to comply.
On the other hand, the position of the European Parliament adds exemptions for research activities and AI components provided under open source licenses, as well as at least one regulatory sandbox (established by public authorities) in every EU member state. Establishing a controlled environment) has been added. Facilitate secure development, testing, and validation for a limited period of time before bringing innovative AI systems to market or entering service according to specific plans under regulatory oversight.
As far as citizens’ rights are concerned, the position of the European Parliament recognizes the right to lodge complaints about AI systems and to receive an explanation for decisions taken on the basis of AI systems that pose a high risk of severely affecting the fundamental rights of citizens. increase.
After the European Parliament adopts its position on AI law, interagency negotiations will take place in a so-called tripartite consultation with the European Council, representatives of European governments and the European Commission. Negotiations will begin on June 21 and an agreement is expected to be reached by November 2023.
The AI law, which is expected to come into force by the beginning of 2024 at the latest, will only apply extraterritorially, as it will also apply to providers and users outside the EU if the output generated by the system is used within the EU. We also anticipate fines of up to €40 million or up to 7% of the company’s global annual turnover for the previous financial year, whichever is higher.
To stay up-to-date on evolving AI legal developments, visit Cooley’s dedicated AI page.
