The European Parliament has passed the groundbreaking Artificial Intelligence (AI) Act, the first comprehensive AI legislation globally, with 523 votes in favor, 46 against and 49 abstentions. The Act, expected to take effect in May, introduces a novel, risk-based approach to AI governance, categorizing AI systems according to potential threats and regulating them accordingly. While the Act faced resistance from some countries concerned about its economic impact, and warnings from major European companies about restrictive regulations, it now covers everyday tools up to more sophisticated healthcare and law enforcement systems.
Notably, the Act bans AI systems capable of cognitive behavioral manipulation, social scoring, and unauthorized biometric identification. A “high-risk” category includes AI in critical infrastructure, educational tools, employment management, and others. These high-risk systems will undergo rigorous pre and post-market assessments. If there are concerns about these systems, the public can raise them with designated authorities. Generative AI platforms are expected to adhere to transparency about functionality and training data in accordance with EU copyright laws, despite not being labeled as high-risk. The AI Act is due to become law by mid-2024, with its provisions implemented in stages over two years. Noncompliance may result in fines of up to 35 million Euros or 7% of global annual turnover.
In addition to AI training and deployment regulations, the Act introduces copyright rules. These include that AI models must reveal summaries of their training data, that disclosures allow creators to recognize if their material was used in training, and that SMEs comply with copyright laws while facing more lenient enforcement. The existing provision for creators excluding their work from use in AI training is maintained.
Once formally adopted by mid-2024, the Act’s provisions will gradually take effect. The EU expects prohibited AI practices or projects to cease within six months, general-purpose AI systems to adhere to new rules after a year, and full enforcement in two years. The AI Act is seen as a landmark moment in technological progress, challenging other regions to speed up their AI governance strategies. This has significant implications for businesses, tech giants, and governments globally, influencing AI landscapes far beyond European borders.
