On 21 May 2024, the Council adopted the AI Act, which establishes a comprehensive regulatory framework for the development and use of artificial intelligence in the European Union. The AI Act classifies AI systems based on the level of risk they pose, with strict requirements for 'high-risk' AI that could threaten people's safety or fundamental rights. These include requirements for rigorous testing, human oversight, and transparency. The legislation also bans certain unacceptable uses of AI, such as social scoring or real-time biometric identification in public spaces.
The adoption of the AI Act follows lengthy negotiations between the European Parliament and the Council as well as extensive discussions with the creative community and civil society organizations. The new rules will now enter into force, and member states will have 24 months to transpose them into national law. The new regulation, which was proposed by the European Commission in 2021, aims to both protect fundamental rights and drive innovation, as Europe seeks to position itself as a global leader in the AI sector.
Experts say there are several factors behind Europe's pioneering role in AI regulation. Raul Rubio Pérez, a legal; expert from Spain, noted the EU's strong commitment to human rights and privacy, though he observed this was more oriented towards safeguarding citizens from private sector abuses rather than government overreach. The EU Commission will have the power to fine companies up to €35 million or 7% of their global annual revenue for breaching the new AI regulations - whichever is higher. This puts major U.S. tech firms firmly in the spotlight, as they have been closely watching the development of this groundbreaking law.
The move has been praised by the publishing industry, with the Federation of European Publishers stating that the "AI Act is a vital piece of legislation that will regulate the role of AI in Europe and help set a global standard for how we expect AI systems to operate."
Before the landmark legislation, UNESCO had adopted the "Recommendation on the Ethics of Artificial Intelligence" in November 2021 laying the road for the latest development. The recommendation, which was approved by all 193 UNESCO member states, set out a comprehensive list of principles and policy actions to ensure AI systems respect human rights, promote societal advancement, and protect the environment.
"Without the right ethical guardrails, AI risks reproducing real-world biases and discrimination, fueling divisions, and threatening fundamental human freedoms" Gabriela Ramos, UNESCO's Assistant Director-General for Social and Human Sciences said back then. At the heart of the UNESCO framework were 10 core principles, including the need for proportionality and "do no harm", the protection of privacy and data rights, and ensuring transparency, accountability and human oversight of AI systems. The recommendation also outlined 11 key policy action areas for member states that ranged from data governance and environmental sustainability to gender equality and public awareness-raising.
While some countries have taken a "light-touch" approach, the European Union has emerged as a global leader in AI governance with its landmark "AI Act" legislation that fulfils and expands UNESCO's AI ethics framework. The new regulation took a tiered approach, applying stricter rules to AI systems deemed to pose higher risks to individuals and society. This included outright banning the use of AI for "unacceptable" applications such as social scoring, predictive policing, and undiscriminating facial recognition.
High-risk AI systems, which could significantly impact health, safety, fundamental rights, the environment, democracy, or the rule of law, were subject to a range of stringent obligations. These included mandatory fundamental rights impact assessments, conformity evaluations, and the implementation of robust risk management and quality control measures. Notably, the regulation also introduced specific provisions for "foundation models" - highly capable AI systems designed to generate content. These models were required to meet enhanced transparency and accountability requirements. To support implementation, the EU established a public database of high-risk AI systems, allowing for greater scrutiny and oversight of these technologies.
The new rules were praised by industry groups, with the Federation of European Publishers stating that the AI Act "will help set a global standard for how we expect AI systems to operate."
While the AI Act has now been approved, it will still take time before the new requirements are fully implemented. Restrictions on advanced "general-purpose" AI systems like ChatGPT won't come into effect until 12 months after the law enters into force, and even then, a 36-month transition period is provided to allow companies to bring their technologies into compliance.
In a statement, the Council highlighted the importance of the AI Act in promoting innovation and public trust in artificial intelligence and said the legislation would make Europe a global standard-setter for trustworthy AI.
The AI Act is seen as a significant milestone in the global efforts to ensure the responsible development and deployment of artificial intelligence technologies. As the first comprehensive legal framework of its kind, it could influence regulatory approaches in other parts of the world.
Visit the website of the EU AI ACT
