The EU agrees to the world's first AI regulation law, the chair of the policy enforcement agency emphasizes that it is the 'world's first AI law'



The European

Commission (EC), the European Union's policy-making body, has announced the world's first bill to regulate artificial intelligence (AI) proposed in the European Union (EU). Tripartite talks between the Council of the EU and the European Parliament , the EU's legislative body, have been underway, and an agreement has finally been reached. The EU's AI regulation law was proposed in April 2021, and adjustments were being made despite fierce protests from European companies .

Artificial Intelligence Act: deal on comprehensive rules for trustworthy AI | News | European Parliament
https://www.europarl.europa.eu/news/en/press-room/20231206IPR15699/artificial-intelligence-act-deal-on-comprehensive-rules-for-trustworthy-ai



EU lawmakers bag late night deal on 'global first' AI rules | TechCrunch
https://techcrunch.com/2023/12/08/eu-ai-act-political-deal/

EU agrees to landmark rules on artificial intelligence | Ars Technica
https://arstechnica.com/information-technology/2023/12/ai-regulation-will-begin-in-the-eu/

From late night on Friday, December 8, 2023 to early morning on Saturday, December 9, 2023, local time, representatives of the Council of the EU, the European Parliament, and the EC held a press conference, and agreed on the AI Regulation Law. I announced what happened. At a press conference, the three parties praised the agreement, saying, ``The agreement on the AI Regulation Act is a landmark achievement after a fierce battle and is historic.''

EC President Ursula von der Leyen also announced the agreement on the AI Regulation Act on her X (formerly Twitter) account. 'This is the world's first AI law. It will be a unique legal framework for the development of trustworthy AI, and it will be a law for the safety and fundamental rights of people and businesses,' Leyen said at X. has made a commitment that incorporates political guidelines and will deliver on it. We welcome political agreement on the AI Act.' Furthermore, when Chairman Leyen assumed the position of EC Chairman, he listed the enactment of an AI regulation law as a priority for his term.



The full details of the AI Regulation Act, which was finally agreed upon in the tripartite talks, will not be revealed until the final bill is compiled and published. Please note that it may take several weeks for the final bill to be published. However, the European Parliament has announced some agreed AI regulation legislation.

At the time of writing this article, the use of AI for the following purposes is completely prohibited.

Use of AI in biometric classification systems that use sensitive characteristics (e.g., political, religious, philosophical beliefs, sexual thoughts, race, etc.)
・Using AI to non-targetedly scrape facial images from the internet or surveillance camera footage to create facial recognition databases.
・Using AI for emotional recognition in the workplace and educational institutions
・Using AI for social scoring based on social behavior and personal characteristics
・Creation of AI systems that manipulate human behavior to circumvent human free will
・Using AI to exploit people's vulnerabilities (age, disability, social or economic situation, etc.)



However, the use of biometric technology in public places by law enforcement will not be completely prohibited, and the use of AI will be allowed for a strictly defined list of crimes. Specifically, the use of AI technology is permitted in 'targeted investigations of persons suspected of committing or convicted of serious crimes.' Specific cases where AI can be used in biometric authentication systems are as follows.

・Victim-targeted searches (kidnapping, human trafficking, sexual exploitation)
・Prevention of specific terrorism threats
- Persons suspected of committing any of the specified crimes listed in the regulations (terrorism, human trafficking, sexual exploitation, murder, kidnapping, rape, armed robbery, participation in a criminal organization, environmental crimes, etc.) identification or location

Additionally, clear obligations will be imposed on AI systems classified as high-risk (those with the potential to cause significant harm to health, safety, fundamental rights, the environment, democracy, and the rule of law). I am. The European Parliament “succeeded in including, among other requirements, a mandatory fundamental rights impact assessment applicable to the insurance and banking sectors”. In addition, ``AI systems used to influence election results and voter behavior'' are also classified as high-risk AI systems. Citizens can file complaints about AI classified as high-risk AI systems and receive explanations about decisions based on high-risk AI systems that affect their rights.



In addition, creators of generative AI applications (general-purpose AI: GPAI) such as ChatGPT and the AI models that support them are required to ``create and publish technical documents related to AI'' and ``comply with EU copyright law.'' They will be required to 'disclose details regarding the content used for training.' It is also a proposal by the European Parliament to require GPAI to comply with transparency requirements.

It should be noted that while commercialized AI models and GPAI are likely to face regulation under the EU's AI Regulation Act, research and development areas are not intended to be subject to the law. Additionally, fully open source AI models have lighter regulatory requirements than their closed source counterparts.

Penalties for violating the AI Regulation Act range from 7.5 million euros (approximately 1.2 billion yen) or 1.5% of sales to 35 million euros (approximately 5.5 billion yen) or 1.5% of sales, depending on the nature of the violation and the size of the company. 7%” will be applied. The agreement also allows for gradual entry into force after the adoption of the law, with six months for prohibition regulations to take effect, 12 months for transparency and governance requirements, and 12 months for other requirements. will be given a grace period of 24 months.

Therefore, the full effect of the EU's AI regulation law may not be felt until 2026, foreign media TechCrunch pointed out.

Related Posts:

in Note, Posted by logu_ii