The first batch of regulations for the EU Artificial Intelligence Act (AI Act) was officially released on the 2nd. After these regulations take effect, EU regulatory bodies will be able to ban AI systems deemed to carry “uncontrollable risks.” Violators will face fines of up to 35 million euros. The official legislation will come into effect on August 1. In response, companies such as Amazon, Google, and OpenAI have signed the agreement, but tech giants like Apple, Meta, and French AI pioneer Mistral have not.

To promote AI innovation, the European Commission will release guidelines on defining AI systems. This aims to help industries determine whether software systems qualify as AI systems. The first wave of regulations, released on the 2nd, will categorize AI systems by risk levels. According to reports, the risk levels are broadly divided into four categories:

  • Minimal risk: No regulation (e.g., email spam filters)
  • Limited risk: Light regulation (e.g., customer service chatbots)
  • High risk: Strict regulation (e.g., AI for healthcare advice)
  • Unacceptable risk applications: Complete ban

The most stringent restrictions on unacceptable risks include AI used for social scoring, AI that manipulates personal decisions subconsciously or deceptively, AI that exploits vulnerabilities based on age, disability, or socioeconomic status, AI predicting crime based on appearance, AI using biometric techniques to infer personal traits (e.g., sexual orientation), AI collecting “real-time” biometric data for law enforcement in public spaces, AI attempting to infer people’s emotions at work or school, and AI using web scraping or security camera footage to create or expand facial recognition databases.

For companies headquartered outside the EU, if they use AI applications within the EU and fail to comply with the regulations, they can be fined up to 35 million euros (approximately 1.19 billion TWD) or 7% of their revenue from the previous fiscal year, whichever is higher. However, fines will not take effect immediately. Companies that have signed the regulations will need to begin complying with the legal procedures, but detailed guidelines and enforcement measures will be released by the authorities in August and take effect at the same time.

Currently, not all companies are on board with the EU AI Act. Since its official launch last September, only about 100 companies have signed, including Amazon, Google, and OpenAI. However, Apple and Meta have not signed, and it is understood that these companies are waiting for the detailed guidelines to be released.

<strong><a href=”https://www.ettoday.net/news/20250204/2902348.htm”> Source: ETtoday News Cloud </a></strong>