The European Union will begin implementing its new regulations on artificial intelligence on August 1.

The European Union has officially published the full text of the EU AI Act in its Official Journal, marking the enforcement starting date as August 1, 20 days after publication. While all provisions will be fully applicable in two years, certain bans and regulations will be implemented sooner.

Implementation of bans on prohibited AI applications, such as social credit ranking systems and real-time emotion recognition in schools and workplaces, will begin in six months. Following that, at the nine-month mark, the EU will introduce codes of practice for AI developers in collaboration with consultancy firms and companies providing high-risk general-purpose models.

One year after publication, makers of general-purpose AI models like ChatGPT will need to adhere to new transparency requirements to ensure user safety and understanding. Additionally, rules regarding generative AI and manipulated media, such as clearly labeling deepfakes and AI-generated content, are included in the EU AI Act.

Furthermore, companies training their AI models must comply with copyright laws unless the model is solely for research purposes. The text of the AI Act specifies that rightsholders can opt out of text and data mining unless it is for scientific research, requiring providers of general-purpose AI models to obtain authorization for such activities.