EU lawmakers recently reached a historic agreement in Brussels, introducing the world's first comprehensive AI act. This pioneering legislation aims to regulate artificial intelligence systems based on their potential risks and impact levels, establishing a unified regulatory framework for the 27 member states. The bill not only promotes innovation but also aims to protect fundamental human rights.

The new bill categorizes AI systems into four risk levels, ranging from unacceptable risk to low risk. According to Reuters, AI applications deemed "unacceptable risk" will be completely prohibited. This includes government-controlled social scoring systems and law enforcement activities that involve real-time biometric monitoring in public spaces. High-risk AI applications, such as those used in critical infrastructure and medical devices, will face strict obligations, including rigorous risk assessments and requirements for high-quality data sets.

The bill also establishes a new European Artificial Intelligence Office, responsible for overseeing the implementation of these regulations. The agency will work alongside the European Commission to ensure consistent implementation of national regulations. Companies that violate the AI act will face fines of up to 35 million euros or 7% of their global revenue, a strong enforcement mechanism aimed at ensuring large technology companies take the regulations seriously.

An important breakthrough in the negotiations was the regulation of powerful general AI models, such as the system behind ChatGPT. Lawmakers agreed to adopt a two-tier regulatory approach. All general AI models must comply with basic transparency requirements, while powerful models deemed to pose "systemic risks" must take on additional obligations, including mandatory model assessments and incident reporting.

The new regulations are expected to have significant global implications. EU regulations, often referred to as the "Brussels effect," frequently become de facto global standards. This act may influence the approaches of other major economies in AI governance.

Key Points:

🌐 ** First Comprehensive AI Act Globally **: The EU introduces a historic AI act, establishing a unified regulatory framework.

🚫 ** Prohibition of Unacceptable Risk AI **: Certain manipulative applications, such as social scoring systems, will be banned.

💰 ** High Fines Ensure Compliance **: Companies violating the rules may face fines of up to 35 million euros or 7% of their global revenue.