The European Union is putting the brakes on the use of artificial intelligence through the only law of its kind in the world.
This article details new measures adopted by the European Union to regulate the use of artificial intelligence, highlighting the importance of protecting citizens' rights and maintaining security while encouraging AI innovation. In Strasbourg, MEPs voted on a series of restrictions to limit the manipulation of public opinion through the dissemination of false images. This is the first law in the world to regulate the use of artificial intelligence (AI).
Restrictions on tampering with false images
Information, text or images generated by artificial intelligence must be clearly labeled. Authorities will only be able to use biometric facial recognition systems to prevent terrorism or search for missing persons.
AI-generated fake images, known as "deep fakes," can manipulate public opinion and have become weapons in political battles. The emergence of a virtual assistant in 2021, trained on a huge amount of data, highlighted both the potential and the risks of this technology.
Notable incidents
Recently, in Hong Kong, an employee of a company was tricked into transferring $25 million into the account of fraudsters at the request of avatars that looked like his bosses in a video phone meeting. Victor Kapra, tech specialist, points out that "AI systems can read in privacy what diseases we have and what biological age we have, and insurance companies could make decisions based on this intimate data".
Measures to prevent abuses
To stop abuses, the European Parliament passed the world's first law regulating the use of AI. Biometric systems, such as facial recognition cameras in public spaces, can only be used by authorities in extreme, clearly regulated cases, such as searching for a missing person or preventing a terrorist attack.
Law enforcement will need a court order to use biometric systems after a crime has been committed. The use of AI for emotion recognition in the workplace and schools is prohibited. Products and services that manipulate human behavior or exploit people's vulnerabilities are also prohibited.
Additional restrictions
No social scores are allowed, similar to the Chinese system, i.e. grades for social behavior. Toys that use artificial intelligence are also restricted. Bogdan Manolea from the Association for Technology and the Internet points out that in the case of children, the reduced ability to discern requires increased attention in the presentation of these smart toys.
Implementation of the new rules
The new rules will come into force by the end of the year. An office for the management of artificial intelligence will be established within the European Commission. MEP Dragoș Tudorache states that "the authorities have two years to prepare their people to apply this legislation. It will be important to find the appropriate human resource for the role that the national authorities will play".
Penalties for non-compliance with the law
Companies that do not comply with the rules can pay fines of up to 7% of turnover, with a ceiling of €35 million for the most serious offences.
Additional details about the law:
The law approved by the European Parliament (EP) aims to respect human rights and maintain security in the field of Artificial Intelligence (AI). The EU is the first entity to adopt such a normative act, which according to the initiators, leaves room for innovation in this industry.
The law prohibits certain AI applications that threaten citizens' rights, including biometric classification systems or mining images from the internet or CCTV footage to create facial recognition databases.
The new provisions prohibit the use of AI applications for emotion recognition at work and in schools, social assessment, certain police activities, but also applications that manipulate human behavior or exploit people's vulnerabilities.
Restrictions on Biometric Identification
The use of biometric identification systems by law enforcement authorities is prohibited in principle, except in situations clearly specified in the law and strictly defined. "Real-time" biometric identification of people can only be applied if strict guarantees are met, for example, its use is limited in time and can only target a certain geographical space. In addition, it needs a judicial or administrative authorization.
Permitted uses include, for example, efforts to find missing persons or prevent a terrorist attack. Post-event use of biometric identification also requires judicial authorization.
Obligations for risk-based AI systems
The law provides clear obligations for other AI systems that have a high degree of risk to health, security, fundamental rights, the environment, democracy and the rule of law. Among the examples of high-risk AI uses mentioned in the law are those in areas such as infrastructure, education, employment, essential public and private services, migration and border management, justice and democratic processes, for example influencing elections.
Citizens' rights
Citizens will have the right to lodge complaints about AI systems and receive explanations for decisions based on high-risk AI systems that could affect their rights.
"Today is again a historic day on our long road to AI regulation," said Brando Benifei, the Italian MEP who promoted the text in the EP together with the Romanian MEP, Dragoș Tudorache. "This is the world's first regulation that sets a clear path to the safe and human-centered development of Artificial Intelligence," he said.
To enter into force, the Law adopted by the European Parliament must also be approved by the European Council, an institution made up of the heads of state and government of the EU member states. Approval will probably take place in May, according to EU decision-making procedures. The law is to be fully applied 24 months after its entry into force, with some exceptions:
- Prohibitions on Prohibited Practices will apply six months after the effective date.
- The codes of practice will apply nine months after they come into force.
- AI rules of general use, including governance, will apply 12 months after entry into force.
- Obligations for high-risk systems will apply at 36 months.
Reactions from the technology sector
Technology players have received the law with caution. "Many of these new rules remain unclear and could slow down the development and implementation of innovative applications," said Boniface de Champris, head of Europe at the Brussels-based CCIA (Computer and Communications Industry Association), an industry lobby representative.
For their part, lobbying organizations report that many details of the AI Act remain open and need to be clarified, for example in terms of standards or transparency obligations. The same organizations said firms such as French start-up Mistral AI and Germany's Aleph Alpha, or tech giants Google and Microsoft, had strong influence during negotiations on the law's provisions.
The lobby was also recognized by the initiator Dragoș Tudorache, who emphasized that the law is balanced and protects the interests of European citizens, leaving room for the development of the sector.
The law passed in the European Parliament comes amid growing global fears that the very rapid development of the Artificial Intelligence sector poses significant risks to human rights and democratic systems. In this context, the European Union aims to establish a balance between the protection of citizens and the promotion of innovation, being a global example in the regulation of an emerging and complex technology.
Comments