EU AI law to ban AI applications takes effect in February

• An AI system that uses manipulative and deceptive techniques, either unconsciously or intentionally, to influence people’s behavior to make them make decisions they would not otherwise have made.
• AI applications that negatively influence human behavior by exploiting human vulnerabilities due to age, disability, or specific social or economic circumstances.
• AI that judges people based on social behavior or known, inferred, or predicted personal or personality traits, resulting in disadvantage or unfavorable treatment.
• An AI system that predicts someone’s risk of committing a crime and bases its judgment solely on profiling of a natural person or assessment of personality traits and characteristics.
• AI applications that create or expand facial recognition databases by untargeted reading of facial images from the Internet or video surveillance recordings
• AI systems for inferring the emotions of natural people in the workplace and educational institutions (exceptions: for medical or safety reasons).
• The use of biometric classification systems to classify individual natural persons based on biometric data to draw conclusions about race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation.
• AI-enabled real-time biometric remote identification system in public places.

However, with regard to the last two prohibitions, it should be noted that EU AI law allows exceptions for law enforcement purposes. Therefore, if you want to use it for this purpose, it is best to carefully evaluate it on a case-by-case basis.

Although the rules for banned AI systems are relatively clearly defined, there are still many loopholes in other aspects. And binding rules and regulations for general purpose AI (GPAI) will also apply from August 2, 2025.

EU AI law defines GPAI as AI applications that use LLMs such as GPT-4o, Llama, etc. Therefore, the regulations that will apply from August are expected to mainly affect major AI companies such as Microsoft, Google, Meta, and OpenAI.