First big breakthrough on the regulation of artificial intelligence. The European Parliament approved the first text (called To the Acts) which defines the rules for artificial intelligence. It is the first global regulation on the subject. The text is passed with 499 votes in favor 28 against and 93 abstentions surpassing even theopposition from the EPP which had presented two amendments to modify part of the measure. Thanks to the Ai Act, technologies such as ChatGPT are imprinted with respect for EU laws and fundamental values and to combat digital discrimination and prevent disinformation and the use of "deep fakes".
"We are writing history! I am proud that the European Parliament has just passed the world's first act on artificial intelligence. A balanced and human-centred legislation that will set the global standard for years to come. Europe will continue to lead in innovation” the president of the Eurochamber wrote on Twitter, Robert Metsol.
Now it opens road to trilogue i.e. the negotiation between the three EU institutions: Parliament, the Council (Member States) and the EU Commission. The first negotiations will start immediately this evening “We are organizing the first trialogue tonight in Strasbourg” commented the EU Commissioner for the Internal Market Thierry Breton: “AI raises many questions from a social, ethical and economic point of view. But this is not the time to hit any “pause button”. On the contrary, it is about acting quickly and taking responsibility."
What is the Ai Act
The AI Act is a pioneering regulation which should apply uniformly in all EU-containing Member States 85 articles and attachments. It is the first legislative text that aims to comprehensively regulate the field of AI. There first draft EU law dates back to 2021 and has since been enriched with further amendments. The purpose of the regulation is to ensure that the technology "complies with the rights and values of the Union, for example in the fields of human supervision, security, privacy, transparency, non-discrimination and social and environmental well-being".
One of the main features of the AI Act is the breakdown of AI applications based on associated risk levels.
The European Commission has identified four levels of risk:
- Unacceptable risk: This level refers to systems that violate the fundamental values of the European Union and will not be allowed within the borders of the EU.
- High risk: Concerns those systems that have a controversial impact on the safety and rights of people. While not prohibiting their use, the EU requires certain requirements to be met to ensure safety and the protection of rights.
- Limited Risk: This category includes systems that do not pose a significant risk and must meet a limited set of requirements, including transparency in the use of artificial intelligence.
- Minimum Risk: There are no specific legal obligations in this category.
Ai Act: what is forbidden
The AI Act prohibits the use of artificial intelligence forremote biometric identification in real time and retrospectively in public spaces, as well as for categorization based on sensitive characteristics. Also prohibited are predictive policing systems, emotion recognition systems andindiscriminate extraction of biometric data from the Internet or from security cameras to create facial recognition databases. In addition, artificial intelligence systems used to influence elections and recommendation systems of social media platforms have been added to the list of high risks. These measures aim to protect the human rights and privacy of citizens.
Ai Act: the sanctions envisaged
More attention is given to categories that can significantly influence people's lives, such as job hunting and education. In these areas, transparency and impartiality requirements of the algorithms used are required. For those who violate the regulation there are provisions severe penalties. expected fines of up to 30 million euros or up to 6% of turnover company year.
Ai Act: Effective no earlier than 2025
The AI Act is expected to come into force no earlier than 2025, also to allow companies to adapt to new laws. In the meantime, to fill this legislative vacuum, the European Union has recently signed a joint statement with the United States to arrive at a common code of conduct on the use of artificial intelligence. “The European Union and the United States reaffirm their commitment to a risk-based approach to AI, to advance trusted and accountable AI technologies. Cooperation on our approaches is essential to promote responsible innovation, which respects rights and security and ensures that AI delivers benefits in line with our shared democratic values” reads the joint document signed during the ministerial meeting of the Council EU-US Trade and Technology held in Lulea, Sweden.
Pending the law, the EU has also proposed to companies that are part of the EU Code of Conduct to introduce a sticker to identify AI-generated content to counter misinformation.
European Union pioneer in regulating AI
The race to regulate artificial intelligence started after the alarm raised by leading AI experts. The European Union, however, had moved for some time to arrive at a text of law and anticipated the United Kingdom and the United States in reaching an agreement on the first law on AI. The United States, despite the words of concern from the creator of ChatGPT, Sam Altman, in front of the US senators, has not yet taken any action to discipline the sector. London, instead, is pointing to atrue leadership in the sector trying to present itself as the seat of aInternational Regulatory Authority on AI. It was a few days ago that the announcement will host the first global summit on the security of artificial intelligence in the fall of 2023.
