A special committee of the European Parliament recently approved the initial version of a bill that aims to aims to establish guidelines for the development and use of artificial intelligence systems in the Continent European.
The approval of this preliminary phase marks a significant step forward in the pursuit of regulations designed to control the rapid advancement of AI technologies. These laws are necessary due to the rapid evolution of these technologies.
see more
Threat to Netflix: Google's free service competes with streaming
Google develops AI tool to help journalists in…
The new legislation, entitled “European Artificial Intelligence Act”, is a first in the West, as in the China and other Asian countries already have regulations that generative AI companies such as OpenAI with your ChatGPT, must follow.
European regulation for artificial intelligence takes an approach based on the risk potential of each AI system. The greater the danger they pose, the more strictly they will be controlled.
The rules also set out specific requirements to be followed by developers, known as “foundation models”.
The risk categories of artificial intelligences
The European Artificial Intelligence Act also categorizes AIs by their risk levels. They are: minimal or no risk, limited risk, high risk and unacceptable risk.
Even high-risk AI can work under regulation, but tools categorized as having unacceptable risk cannot even work in Europe. Unacceptable risk AIs are:
- Systems that use subliminal, manipulative, or deceptive techniques to distort their own behavior;
- Artificial intelligences that exploit vulnerabilities of specific individuals or groups;
- Biometric categorization systems based on sensitive attributes or characteristics;
- AIs used for social scoring or reliability assessment;
- AI systems used for risk assessments that predict criminal or administrative offences;
- Systems that create or expand facial recognition databases through non-targeted scraping;
- Mechanisms that infer emotions in law enforcement, border management, the workplace, and education.
To prevent the proliferation of harmful artificial intelligences, such as high risk and unacceptable risk, the new European Artificial Intelligence Law will force developers to follow standards predetermined.
One of the aspects of this database is the prior testing of the artificial intelligences, which must occur in all cases, under penalty of sanction to the developer who refuses to carry out the tests.
“Providers of these AI models will be required to take measures to assess and mitigate risks to fundamental rights, health, safety, environment, democracy and the rule of law,” said Ceyhun Pehlivan, a European lawyer interviewed by the broadcaster CNBC.
In a word, experts in Artificial Intelligence
As it couldn't be otherwise, the news that a law is being created to regulate AI in Europe fell like a bomb among specialists in the field and some technology industries.
In advance, it is worth noting that the European Artificial Intelligence Act is still being developed and that the recent approval is only preliminary. Even so, some important actors express concern.
According to the Computer and Communications Industry Association of Europe, the scope of the new law has been extended too far, which could cause it to “catch harmless AIs”.
“It is troubling to see that large categories of useful AI applications – which pose very limited, if any, risks – may now face strict requirements, or they may even be banned from Europe," Boniface de Champris, policy manager at the organization, said in a statement sent to CNBC. by email.
According to Dessi Savova, head of technology at the renowned law firm Clifford Chance, said the AI control laws emerging in Europe will influence the world.
“The reach of the proposed AI rules in Europe will reach AI developers around the world,” she said.
“However, the central question is whether the European Artificial Intelligence Act will set a standard for AI in the world. That's because China, the US and the UK, to name a few, are defining their own AI policy and regulatory approaches,” questions Savova.
“But, undeniably, they will all closely follow the negotiations that are taking place in Europe to adapt their own approaches”, she concluded.
Finally, Sarah Chander, Senior Policy Adviser at European Digital Rights, highlighted that new European AI laws will require developers to demonstrate their “power”.
“While these transparency requirements [contained in the bill] do not eradicate concerns about infrastructure and economics, with the development of these vast AI systems, they require technology companies to disclose the amounts of computational power needed to develop them”, he recalled, in conversation with the CNBC.
So far, the next steps of the European Artificial Intelligence Act have not yet been released by the European Parliament.
Graduated in History and Human Resources Technology. Passionate about writing, today he lives the dream of acting professionally as a Content Writer for the Web, writing articles in different niches and different formats.