EU Passes World's First 'AI Act'...Up to 7% Annual Revenue Fine

The European Union (EU) Parliament has passed the world's first comprehensive artificial intelligence (AI) regulatory bill.

On the 13th (local time), the European Parliament announced that the final draft of the 'AI Act' was approved at the plenary session held in Strasbourg, France, with 523 votes in favor, 46 against, and 49 abstentions.

Thierry Breton, EU Commissioner for Internal Market, said in a statement, "Europe is now the global standard-setter for trustworthy AI."

Accordingly, the AI Act will come into effect after publication in the official journal once the ministers of the 27 EU member states give final approval next month. Some prohibited provisions will apply six months after enforcement and will be introduced gradually, with full implementation expected after 2026.

The EU divides AI application areas into four risk levels and regulates them differentially. For high-risk categories such as healthcare, education, public services, elections, critical infrastructure, and autonomous driving, AI use must be supervised by humans. A risk management system must also be established.

Companies developing Artificial General Intelligence (AGI) are subject to a 'transparency obligation.' They must comply with EU copyright law and disclose the content used during AI training. Measures must be taken to prevent incidents defined by the EU as systemic risks, such as widespread cyberattacks. Obligations for information disclosure and notification are also imposed, and the use of certain technologies, such as real-time remote biometric identification systems using AI, is prohibited.

Violations of the law may result in fines of up to 7% of global revenue.