Rules from the European Union to regulate artificial intelligence
European Union (EU) institutions have reached an agreement on a new law that will impose strict rules on various artificial intelligence technologies such as ChatGPT and Gemini.
AA
The Council of the EU announced that an agreement has been reached on the "Artificial Intelligence Act," which was negotiated between member states and the European Parliament (EP) and will introduce the world's first comprehensive rules for artificial intelligence. Accordingly, artificial intelligence systems to be used in European countries will be required to be safe and to show respect for fundamental rights and Union values.
Artificial intelligence systems will be regulated with a "risk-based" approach according to their potential to cause harm to society.
Stricter rules will be introduced for high-risk artificial intelligence systems. Clear criteria will be established to distinguish artificial intelligence from simple software systems.
It will be ensured that artificial intelligence does not, under any circumstances, affect the competencies of member states regarding national security or any body tasked in this field.
The artificial intelligence law will not apply to systems used solely for military or defense purposes.
Artificial intelligence systems that present limited risk will clearly notify users that the content is generated by artificial intelligence so that they can make informed decisions. These systems will be subject to light transparency obligations. Various high-risk artificial intelligence systems will be permitted. However, the access of high-risk artificial intelligence systems to the EU market will be subject to a series of rules and obligations.
SOME BANS WILL ALSO BE IMPOSED ON ARTIFICIAL INTELLIGENCE
The risk in some uses of artificial intelligence will be considered unacceptable, and these systems will be banned in the EU.
Cognitive behavioral manipulation, untargeted scraping of facial images from the internet or closed-circuit camera system footage, emotion recognition in the workplace and educational institutions, social scoring, and biometric classification to extract sensitive data such as sexual orientation or religious beliefs will fall under the scope of the ban.
Law enforcement agencies will be able to use artificial intelligence in their activities. Law enforcement agencies will be able to deploy a high-risk artificial intelligence tool that would not normally pass the conformity assessment procedure through an emergency procedure.
Security units will be able to use real-time remote biometric identification systems in public spaces with permission in exceptional and necessary cases.
The use of such artificial intelligence systems will be limited to situations such as terrorist attacks, the prevention of existing or foreseeable threats, and the search for persons suspected of the most serious crimes. However, a special mechanism will be implemented to ensure that fundamental rights are adequately protected against potential abuses of artificial intelligence systems.
Special rules will be introduced for large systems capable of performing a wide variety of different tasks, such as generating video, text, images, speaking in another language, calculating, or writing computer code. It will be ensured that the general-purpose artificial intelligence systems in question comply with various transparency obligations before being released to the market.
An "Artificial Intelligence Office" will be established within the EU Commission, tasked with supervising advanced artificial intelligence models, contributing to the development of standards and testing practices, and implementing common rules in all member states.
A scientific panel consisting of independent experts will provide advice to the Artificial Intelligence Office. The Artificial Intelligence Board, which will consist of representatives of member states, will work as a coordination platform and advisory body for the Commission. Fines for violations of the artificial intelligence law will be calculated based on the higher of either a percentage of the company's global annual turnover in the previous financial year or pre-determined amounts.
In case of violation of banned artificial intelligence applications, a fine of 7 percent of the company's turnover or 35 million euros, in case of non-compliance with obligations, 3 percent of its turnover or 15 million euros, and in case of sharing incorrect data, 1.5 percent of its turnover or 7.5 million euros will be imposed. The value that is higher for the company will be used when imposing the fine.
TECHNOLOGY COMPANIES WILL ALSO BE EXPECTED TO COMPLY WITH THESE RULES
The law, which will enter into force after being officially approved by EU countries and the EP, will begin to be applied 2 years after official approval. The EU Commission had prepared the first legislative proposal containing the framework of new rules on artificial intelligence 2 years ago and submitted it to the member states and the EP.
This proposal introduced a number of limitations and transparency rules in the use of artificial intelligence systems. In the Commission's legislative proposal, artificial intelligence systems were divided into 4 main groups: unacceptable risk, high risk, limited risk, and minimal risk.
With the new law, various artificial intelligence technologies such as Google's artificial intelligence model Gemini and ChatGPT will be required to comply with the new rules.