EU AI Act: Prohibited AI Ban and AI Literacy Rules Now in Force
The EU AI Act has officially come into effect, marking a significant milestone in the regulation of artificial intelligence within the European Union. With the passing of the first major compliance deadline on February 2, 2025, regulators have implemented a ban on certain high-risk AI applications. Companies developing or deploying AI systems must now ensure that their staff possesses a sufficient level of AI literacy.
Prohibited AI Practices
According to Article 5 of the EU AI Act, specific AI applications that pose unacceptable risks to fundamental rights and core Union values are banned. The law prohibits the sale, deployment, or use of any AI system that:
- Uses subliminal, manipulative, or deceptive techniques that distort a person’s behavior or impair their decision-making;
- Exploits the vulnerabilities of individuals based on age, disability, or socio-economic status;
- Evaluates individuals based on their social behavior or personality traits, leading to unfair treatment;
- Predicts the risk of criminal offenses based solely on profiling;
- Creates facial recognition databases through untargeted scraping of facial images;
- Infers emotions in individuals at work or in educational settings;
- Uses biometric categorization to infer sensitive demographic characteristics;
- Collects “real-time” biometric data in public spaces for law enforcement purposes.
While there is an exemption for the use of AI systems designed to infer emotions in the workplace for medical or safety purposes, this reflects a critical balance between AI innovation and the protection of consumer rights.
Guidance and Clarifications
The European Commission has released Guidelines that clarify the scope of the ban and provide practical examples to aid understanding of the Act’s applicability. Notably, the term ‘placing on the market’ refers to any availability of an AI system, regardless of the means of supply, including APIs and cloud services.
The ‘putting into service’ aspect covers the supply of AI systems for first use, while ‘use’ encompasses the deployment of an AI system throughout its lifecycle. Importantly, misuse of an AI system falls under the prohibited practices as well.
For instance, manipulative practices by AI systems are prohibited if they fulfill certain conditions: they must constitute placing on the market, involve deceptive techniques, materially distort behavior, and likely cause significant harm to the individual.
AI Literacy Requirements
Effective February 2, 2025, the AI literacy provisions under Article 4 took effect, mandating that all providers and deployers of AI systems ensure their personnel possess a sufficient level of AI literacy. This encompasses the necessary skills, knowledge, and understanding required to make informed decisions regarding AI deployment.
The European Artificial Intelligence Office has established a Living Repository of AI Literacy Practices to promote awareness and education regarding AI literacy.
Next Steps for Organizations
Organizations must first assess whether they are using any AI applications prohibited under Article 5. If so, they should engage stakeholders to phase out these AI systems and discontinue their use. Establishing procedures for identifying future AI initiatives that may intersect with the ban is advisable, alongside implementing employee training to ensure compliance across teams.
Even if specific prohibitions do not apply, the AI literacy requirements span a broader range of AI activities, offering an opportunity to develop comprehensive AI governance, training programs, and safeguards against potential misuse.
This regulatory framework aims to ensure that while AI technology advances, it does so in a way that prioritizes ethical considerations and consumer protection.
