The European Union's (EU) push for harmonised regulation of artificial intelligence (AI) systems entered a new chapter when the European parliament and the Council recently proposed Regulations that harmonised rules on artificial intelligence. This Proposal is the first worldwide initiative that addresses AI implementation concerns related to the degree of autonomy AI systems have in decision-making and the capacity to control their processes. In respect to a successful digital transformation, the EU’s main goal is to boost innovation and business activity while ensuring that people's trust in new technologies is maintained.

AI enablement issues

AI is a system comprised of three main interconnected components: data processing, machine learning and business implementation. Algorithms define the concise process of an AI workflow. AI processes training data and forms patterns, but algorithms are the decision-makers that set the criteria for action. Since AI is a self-learning system, the underlying issue is the limited control of the way data is processed and on the outcome of the algorithm. Therefore, it is difficult to question a decision an AI system has made based on variable data patterns and automated processes.

Experts in the field have reported instances where AI perpetuates discrimination due to bias in initial data (e.g. AI rating male co-workers as more efficient at work than female colleagues can lead to unfair treatment in the workspace for businesses implementing AI technologies). Such techniques have an innate tendency to amplify biases present in data regarding factors such as race, gender, and education. These practices are proven to be diminishing and in the long run could lead to issues like discrimination in the workplace, ESG violations and costly expenses for businesses.

Liability distribution is another problem that could arise given the significant role of AI in our lives. The autonomy of AI systems makes it difficult to pinpoint the responsible party when breaches of law are identified. Hence, if a person suffers harm or when property is damaged as the result of an algorithm's decision, obtaining compensation may be difficult. Hence, adjustments to liability regimes are without a doubt necessary.

AI policy makers

The EU has demonstrated its willingness to address the above issues in AI legislation by setting special legislative bodies and policy makers. The European Parliament has appointed a special Committee on Artificial Intelligence in a Digital Age, which at present is considering the future impact of the Proposal and ways to improve regulatory measures.

Furthermore, the European Data Protection Supervisor (EDPS) has welcomed the Proposal and shown a readiness to fulfil its new role as the AI regulator for EU's public administration. The EDPS has already addressed the need to ban the use of remote biometric identification systems in public spaces, which supplemented by AI technologies might pose high risk of intrusion in privacy and violation of fundamental human rights.

Bulgaria's AI initiative

In line with European values for holistic digital transformation, the Bulgarian Academy of Sciences initiated a plan for the "Development of Artificial Intelligence in Bulgaria by 2030, which the Bulgarian Ministry of Transport, Information Technology and Communications later adopted as a national strategy for the development of AI. The Plan recognises EU priorities, including those for the protection of fundamental human rights and consumer rights in relation to high-risk AI, while at the same time establishing the groundwork for the development of a regulatory framework for reliable AI systems in Bulgaria. The Plan also lays out new concepts such as an AI "trust ecosystem" and a national framework on risk assessment of AI systems, which is intended to prevent unethical use. All this illustrates Bulgaria's ambition to balance regulatory measures and position itself as the next hub for AI development.

The impact of AI regulation

The EU is seeking to embrace regulatory mechanisms at the early age of AI development, so the advancement of such technologies has minimal adverse effect on society and the block's economy. In addition, taking on such measures at a time when AI technologies are not widely implemented guarantees avoidance of regulatory fragmentation across the EU. Some other EU members have also started to raise initiatives examining AI problematics such as Germany's Institute for Ethics in Artificial Intelligence.

The coordination of AI policies will serve human-centered technology, while taking advantage of its vast implementation across industries, and trigger an efficiency boost for the economy. Avoiding scenarios of irreversible fallout when enabling AI technologies is at present one of EU's top legislative priorities in the context of the wider digital transformation initiative.

In the near future, the EU plans to adopt legal frameworks consolidating ethical standards and civil liability in the context of AI since its use will be further expanding. 

Co-authored by Elitsa Hinova, Trainee Lawyer at CMS Sofia