The European Union is so far the only judicial jurisdiction in the world to advance comprehensive rules forward for artificial intelligence through the AI law.
Jack Silva Norfuto Gety pictures
The European Union officially began to enforce the law of artificial intelligence, the teacher on Sunday, which paves the way for difficult restrictions and possible possibilities of violations.
The European Union Law of Amnesty International, the first organizational framework of its kind of technology, was officially entered into in August 2024.
On Sunday, the deadline for the ban on some systems and requirements of artificial intelligence was set to ensure sufficient technological literacy between the pillars.
This means that companies must now comply with restrictions and can face penalties if they fail to do so.
Artificial Intelligence Law prohibits certain applications of artificial intelligence, which it considers an “unacceptable danger” for citizens.
These include social registration systems, actual time recognition and other forms of biological definition that classify people according to race, sexual life, sexual orientation, other features, and AI's “manipulation” tools.
Companies Facial fines Of up to 35 million euros (35.8 million dollars) or 7 % of its global annual revenues – whatever the amount is higher – for violations of the European Union Law.
The size of the sanctions depends on the violation and size of the company that has been fined.
This is higher than possible fines under gross domestic productSigher Digital Privacy Law in Europe. Companies face fines of up to 20 million euros or 4 % of the annual world rotation of GDP violations.
“It is not perfect” but “affects the need for it”
It should be noted that the artificial intelligence law is still in effect – this is the first step in a series of many upcoming developments.
Tasos Stampelos, head of the European Union’s general policy and government relations at Mozilla, previously told CNBC that although it is “not perfect”, the European Union's artificial intelligence law is very required.
“It is extremely important to realize that the artificial intelligence law is often legislating product safety,” Steplasses said in a CNBC committee in November.
“With the rules of the safety of the product, the moment when it is in place, it is not an end. There are many things coming and follow -up after adopting an act,” he said.
Stampelos added: “At the present time, compliance depends on how the standards, guidelines, secondary legislation, or derivative tools that follow the artificial intelligence law, which will actually lead to a provision of compliance.”
In December, the UEFA Office of Amnesty International, a newly created body that regulates the use of models in accordance with the AI law, has published a second -class AI's practices blog for general purposes, which indicates systems such as the Openai GPT family from large language models , Or llms.
The second draft contained exemptions for the providers of some open source artificial intelligence models, with the requirements of the “regular” GPAi models to be subjected to strict risk assessments.
Place the global standard?
Many executives and technological investors are not satisfied with some of the most cushioned aspects of the artificial intelligence law and anxiety that it may suffocate innovation.
In June 2024, Prince Constantine of the Netherlands CNBC told an interview that he was “truly concerned” about Europe's focus on artificial intelligence organization.
“Our ambition appears to be limited to being a good organizer,” said Constongen. “It is good to have handrails. We want to bring clarity to the market, the ability to predict and all of this. But it is very difficult to do this in such a fast space.”
However, some believe that the presence of clear rules for Amnesty International can give the leadership of Europe.
“While the United States and China are competing to build the largest models of Amnesty International, Europe offers leadership in building the most worthy of confidence,” said Diane Bogdanov, Director of Engineering and Growth Intelligence at the Bulgarian Payhawk Company.
“The requirements of the European Union Law of Amnesty International on the discovery of bias, the evaluation of regular risks, and human control, do not limit innovation – They define what seems good. “
