The European Union's Artificial Intelligence (AI) Act, which took effect on August 1, requires AI developers to comply with strict measures and imposes heavy fines for violations. violation.
The European Union has published the final text of the Artificial Intelligence (AI) Act, marking the world's first comprehensive set of AI rules. After being approved by the EU Council in May 2024, the Act will be implemented from August 1, 2024. The law aims to minimize risks from ‘high-risk’ AI applications, protecting democracy, human rights, the environment and the rule of law.
However, the new AI regulations pose significant challenges for companies around the world, especially those in China. Patrick Tu, co-founder and CEO of Hong Kong-based Dayta AI, predicts operating costs will increase by 20% to 40% to comply with the EU’s strict rules, indicating a broader financial and operational impact for global AI developers.
The AI Act requires EU-based AI developers to comply with its regulations by August 2, 2026. Each EU member state must create at least a regulatory sandbox for AI technology, providing an environment for developers to test AI systems within legal limits without hindering technological development.
For developers, vendors, and users of biometric AI, the deadline is February 2, 2025. The law prohibits AI applications that pose “unacceptable risks,” including biometric classification based on sensitive characteristics, emotion recognition in workplaces and schools, and the mass collection of facial images for facial recognition databases. Notably, there are specific exemptions for law enforcement under certain conditions.
By May 2, 2025, the newly established Office of AI will issue codes of practice to guide AI vendors on compliance. General-purpose AI systems, such as ChatGPT, must comply with copyright and transparency standards starting in August 2025. Developers of high-risk AI systems have until August 27, 2027, to meet additional requirements set out in Annex I of the AI Act.
Penalties and legal implications for violations of the AI Act
The AI Act imposes stiff penalties for non-compliance, ranging from €35 million ($38 million) to 7% of a company’s total revenue in the previous year, whichever is higher. “The EU AI Act is the first major attempt to regulate AI globally,” said Emma Wright of the law firm Harbottle and Lewis. “It remains to be seen whether compliance costs will stifle innovation or whether AI governance models will become a key EU export.”
The rapid development of AI technology, especially generative AI like ChatGPT, has outpaced regulatory efforts. Countries and trade blocs around the world are scrambling to establish legal frameworks to regulate the use of AI, reflecting a growing recognition of the need for AI governance.