The European Union AI Act: Key Literacy and Practice Bans
The European Union AI Act officially came into force on August 1, 2024, marking a significant milestone in the regulation of artificial intelligence within the EU. Its first obligations are set to take effect on February 2, 2025.
Overview of the AI Act
This legislation establishes a uniform legal framework for AI systems, designed to govern their development, marketing, implementation, and use across the EU. It aims to ensure a high level of protection for fundamental rights, as outlined in the Charter of Fundamental Rights of the European Union.
Promoting AI Literacy
One of the primary obligations under the AI Act is to enhance AI literacy among organizations. This requirement ensures that individuals working with AI possess the necessary knowledge and skills to navigate this complex landscape.
While the act does not specify a concrete standard for AI literacy, it mandates that organizations implement training programs and responsible AI policies tailored to their specific needs. The European Commission’s AI Office serves as a valuable resource, providing a repository of non-binding AI literacy practices to assist organizations in compliance.
Prohibited AI Practices
In addition to promoting literacy, the AI Act prohibits certain AI practices that infringe upon fundamental rights. These prohibited practices include:
- AI systems that manipulate individuals through subliminal techniques.
- Exploitation of vulnerabilities related to age or disability.
- Classification of individuals based on social behavior.
- AI-driven emotion detection, except in medical or safety contexts.
- Unauthorized collection of facial recognition data from online sources or CCTV.
Violations of these prohibitions can result in substantial financial penalties, with fines reaching up to €35 million or 7% of a company’s annual turnover.
Ensuring Compliance
To align with the AI Act, employers must evaluate their existing AI systems and classify their associated risk levels. Those systems deemed unacceptable will need to be discontinued. Although not mandatory, drafting an AI policy can provide crucial guidelines on the usage of AI and literacy within an organization.
Looking Ahead
As further obligations under the European Union AI Act come into effect in August 2025, organizations are encouraged to stay informed and seek professional guidance to navigate the evolving landscape of AI regulation.
