The EU Artificial Intelligence Act is a major piece of legislation in the European Union in the field of artificial intelligence, which aims to provide a unified regulatory and legal framework for the development and application of artificial intelligence. The bill was introduced by the European Commission in April 2021 and after several rounds of revisions and consultations between the European Parliament and the Council of the European Union, a final agreement was reached in January 2024. The European Parliament is scheduled to vote on the legislation on March 13, and one lawmaker said he expects it to be easily passed. According to the current timetable, the bill will basically come into force in 2026, and some provisions will come into force this year.
The bill would regulate different uses of AI based on risk, including a ban on certain use cases, such as "emotion recognition" systems in the workplace, which would face the highest fines: 7% of global revenue or 35 million euros, whichever is higher. The EU will not be the first jurisdiction to regulate AI, but the size of the European market means that the EU AI Act will have a global impact.
The core idea of the EU AI Act is that AI should be in line with the values and principles of the EU, respecting human dignity, freedom, democracy, equality, the rule of law and human rights. To that end, the bill will classify and regulate AI applications based on the harm and risks they may cause. This classification is divided into four main categories: unacceptable risk, high-risk system, limited risk, and minimum risk.
Unacceptable risks include AI systems that perform social credit scores on individuals or real-time facial recognition in public places, which are considered serious threats to people's fundamental rights and freedoms and are therefore prohibited or severely restricted. High-risk systems include AI systems that involve human health, safety, or fundamental rights, such as applications in areas such as healthcare, transportation, education, employment, justice, etc., which will be subject to higher quality and transparency requirements, including data quality, traceability of AI systems, possibilities for human intervention, risk management measures, etc. Limited risk systems include AI systems that may influence or manipulate user behavior, such as software that generates tampering** or "deepfake" imagery, which must disclose that this is AI-generated content so that users can make an informed choice. Least-risk systems include other AI systems that do not fall into the above categories, such as spam filters and AI-generated video games, which will not be subject to the Act but will still be subject to existing laws and regulations.
The introduction of the EU Artificial Intelligence Act will have an important impact on the development and governance of global artificial intelligence, and has also aroused the attention and discussion of all parties. Some believe that the bill will provide a clear and acceptable rule for AI innovation and competition, which is conducive to protecting human rights and social well-being. Others are concerned that the bill will impose too many burdens and restrictions on the development and application of AI, which could hinder Europe's progress and competitiveness in the field of AI. Others called for an international regulatory framework for AI to overcome geopolitical and economic divisions and achieve sustainable and responsible AI development.
Information Reference**:
1 Blockbuster Starter |The final version of the EU Artificial Intelligence Law: Chinese translation of the full text (140,000 words) - Shanghai Collaborative Innovation Center for Artificial Intelligence Social Governance.
2 the act texts | eu artificial intelligence act.
3 The EU is poised to develop comprehensive AI rules or have a global impact.
4 the ai act: what you need to know | eu artificial intelligence act.
5 The EU Artificial Intelligence Act is a cause for concern, which will impose strict regulations on high-risk applications.
6 Five Possible Ways to Regulate AI - The New York Times Chinese
7 Artificial Intelligence Act - Encyclopedia
8 EU Artificial Intelligence Act: Risk Classification and Regulatory Measures - Zhihu.
9 EU Artificial Intelligence Act bans applications such as real-time facial recognition, imposes strict regulations on high-risk applications - Artificial Intelligence - CNBETAcom.