What is the Regulation about?

Making artificial intelligence (AI) safe and trustworthy is the goal of the European Artificial Intelligence Act (AI Act), which came into force on 1 August 2024 and will be implemented from 1 August 2026.

Who is affected?

AI is versatile and can be used wherever large volumes of data are processed. Applications can be found across almost all economic sectors and business fields: from common image, speech, and text recognition to smart home applications and autonomous vehicles, as well as industrial applications enabling predictive machine maintenance.

What does the regulation propose?

AI is not equal to AI – the AI Act distinguishes between application areas and four risk categories. Different requirements are tied to these risk categories. AI systems that contradict ethical principles in the EU and present unacceptable risks, such as social scoring systems, are completely prohibited.

Particularly relevant for businesses are high-risk applications. These will be subject to strict requirements in the future. A comprehensive quality and risk management system must be established, documenting and verifying decision-making processes, data quality, and transparency. High-risk systems include AI-based applications used in personnel management, education and training, critical infrastructure, or as safety components or components in industrial settings.

For low-risk applications, such as chatbots, transparency obligations are foreseen. Minimal-risk systems, such as AI-supported video games or spam filters, do not entail additional legal requirements.

Special rules apply to "general-purpose AI systems." These include applications capable of creating texts, images, videos, and more through generative AI, intended for broad applications rather than specific use cases. Besides a code of conduct to be developed, systemic risks for particularly powerful models will be minimized through various regulations.

The Act can be accessed at eur-lex.europa.eu.

An example: AI in personnel selection

CV parsing is software for the automatic analysis and extraction of CV data in the recruitment process. It filters applicant data, such as education details, supporting personnel decision-makers. Such systems will face stricter requirements in the future as AI applications used in recruitment are listed as high-risk systems.

This will mean, for example, that users will have to log processes to enable traceability of personnel selection or provide certain information transparently to applicants.

What is important for businesses now?

The proposed legal framework can contribute to strengthening trust and acceptance of AI. Businesses need clear and understandable criteria to identify specific AI risks swiftly, easily, and securely. Bureaucratic burdens and double regulation should be avoided to minimize strain and ensure European competitiveness and the further development of AI in Europe.