Artificial intelligence, what the Brussels rules provide
With the aim of establishing common standards, yesterday the European Commission presented the new rules for the use of artificial intelligence. Companies that do not comply with the rules will risk penalties of up to 6% of turnover
Artificial intelligence must rhyme with confidence for the EU.
Yesterday, the European Commission presented a package of rules on the use of artificial intelligence ( broadly in line with what the media anticipated last week ), including a ban on most surveillance systems.
This is Brussels' attempt to set global standards for AI, a key technology dominated by China and the United States.
All systems deemed "a clear threat to security, livelihoods and people's rights", such as "social scoring" systems, will be banned. The use of biometric identification systems, such as facial recognition, in public spaces "is prohibited in principle". The exceptions allowed concern the fight against terrorism, the protection of security or the search for a missing child. This is stated in the draft law on AI presented by the EU Commission.
High-risk applications “will be subject to stringent requirements before they can be placed on the market”. In case of derogation, use "is subject to the authorization of a judicial or independent body and to appropriate limits of time and space".
According to the document, companies that do not comply with EU regulations could face fines of up to 6% of their turnover.
The proposals are the EU's latest move to maintain its role as a global standard-bearer for technology regulation, anticipating the US and China.
European Commissioner for the Internal Market Thierry Breton said that Europe will become the first continent to provide guidelines on the use of artificial intelligence.
All the details.
THE CLASSIFICATION OF AI APPLICATIONS
In the package of rules proposed by the European Commission to introduce a legal framework that puts order in the world of AI, the EU executive classifies new technologies by risk levels: unacceptable, high, limited, minimal.
HIGH RISK SYSTEMS
Among the systems considered to be at high risk, and therefore subject to periodic requirements and checks, there are technologies used in the transport sector "which could endanger the life and health of citizens" and assistance mechanisms in school or professional training “Which can determine access to education and career paths”. But also programs for the management of workers and access to work (for example the software for sorting the CVs of candidates for a job).
These include machines "that manipulate human behavior to evade the free will of users", such as games for minors that use voice assistance and are capable of encouraging dangerous behavior, but also programs that assign a social score to citizens in function of their behaviors.
AND THOSE WITH MINIMUM RISK
On the other hand, video games enabled for artificial intelligence and spam filters fall under the category of minimal risk. On these the Commission has chosen not to intervene with new rules since "these AI systems represent only a minimal or no risk to the rights or safety of citizens".
ABOUT FACIAL RECOGNITION
All biometric identification systems are considered to be of high risk and therefore, while not banned outright, they will be subject to a series of requirements to protect privacy.
But their use in real time in public spaces to fight crime will – in principle – be prohibited.
The EU will therefore not become like China, where facial recognition systems, or remote biometric identification, are used on a massive scale.
EXCEPTIONS PROVIDED
However, the European executive admits the use of biometric identification only if “strictly necessary” in exceptional cases. Such as to search for a missing child, to prevent an imminent terrorist threat or to detect, locate, identify or prosecute a person suspected of having committed a serious crime.
AUTHORIZATION FROM GUIDICIAL AUTHORITY REQUIRED
This use of remote biometric recognition techniques must be authorized by the judicial authority, or by another independent institution, and must have appropriate limits in time, space and the databases used.
THE SANCTIONS FOR VIOLATIONS
Companies that violate the rules face fines of up to 6% of their global turnover or 30 million euros, whichever is higher.
PLANNED CREATION OF A EUROPEAN COMMITTEE FOR AI
In terms of governance, the Commission proposes that national market surveillance authorities oversee the new rules. The creation of a European committee for artificial intelligence is also planned, which will facilitate the implementation of the new regulatory standards.
VON DER LEYEN: "A PLAN TO GUARANTEE THE EU LEADER POSITION IN AI"
“Today we present new rules for reliable artificial intelligence. They set high standards based on different levels of risk. Artificial intelligence is a fantastic opportunity for Europe. And citizens deserve technologies they can trust ”. The president of the EU Commission, Ursula von der Leyen, wrote it on Twitter, commenting on the new regulation proposal for AI presented by Brussels.
“Today we are also presenting a coordinated plan to outline the reforms and investments we need to secure our leading position in Ai, globally. With almost € 150 billion in digital investments – 20% of the budget – Next Generation Eu will help strengthen the excellence in EU AI, ”added von der Leyen.
VESTAGER: "TRUST IS NOT OPTIONAL ON ARTIFICIAL INTELLIGENCE"
"As far as artificial intelligence is concerned, trust is not optional, it is indispensable" underlined Commissioner Margrethe Vestager, presenting the new rules on artificial intelligence. "These rules represent a breakthrough, which will allow the EU to lead the development of new global standards to ensure that AI can be trusted." “By setting the rules we can pave the way for ethical technology around the world and ensure that the EU remains competitive. Our rules will be adapted to future needs and in favor of innovation and will intervene where strictly necessary: when the safety and fundamental rights of EU citizens are at stake ”, added the Commissioner.
THE CRITICS OF ACTIVISTS FOR CIVIL RIGHTS
Civil rights groups, however, have warned that loopholes in the proposal, which include hefty fines for violations and set strict safeguards for high-risk applications, could leave room for government abuses of technology.
European Digital Rights highlighted worrying gaps in the proposal. “The bill does not prohibit the full extent of unacceptable uses of AI and in particular all forms of mass biometric surveillance. This leaves a worrying gap for the discriminatory and surveillance technologies used by governments and companies, ”said Sarah Chander of the association as reported by Reuters .
THE NEXT STEPS
The draft of the document is still subject to change.
In fact, the regulatory package will have to obtain the green light from the governments and the EU Parliament before entering into force.
HERE THE FULL DOCUMENT OF THE RULES PROPOSED BY THE EUROPEAN COMMISSION
This is a machine translation from Italian language of a post published on Start Magazine at the URL https://www.startmag.it/innovazione/intelligenza-artificiale-ue-regole/ on Thu, 22 Apr 2021 13:56:08 +0000.