Artificial Intelligence Act: A Comprehensive Overview
The EU’s AI Act represents a landmark initiative as the world’s first legal framework focused on artificial intelligence. Its primary aim is to ensure that AI systems are developed and utilized in a manner that is safe, ethical, and trustworthy.
What is the AI Act?
The AI Act is designed to impose obligations on the providers and deployers of AI technologies, regulating the authorization of AI systems within the EU single market. The legislation seeks to address risks associated with AI, such as bias, discrimination, and accountability gaps, while promoting innovation and encouraging the adoption of AI technologies.
As the first regulatory framework for AI, it has the potential to set a global standard akin to the General Data Protection Regulation (GDPR) has done for data privacy, fostering ethical and trustworthy AI worldwide.
Risk-Based Classification
The AI Act categorizes AI applications based on their risk levels, establishing specific rules for each category:
- Minimal or No Risks: Most AI systems, such as AI-powered games or spam filters, are deemed low-risk and are not regulated under the AI Act.
- Limited Risks: AI systems like chatbots must adhere to transparency obligations, such as informing users when content is generated by AI.
- High Risks: High-risk systems, including those used in medical diagnostics or autonomous driving, must comply with strict requirements, including rigorous testing and human supervision.
- Unacceptable Risks: Systems that threaten safety, rights, or livelihoods, such as predictive policing or real-time facial recognition, are banned within the EU.
General-Purpose AI (GPAI)
General-Purpose AI models are designed to perform a variety of tasks, such as text generation and image recognition. Those that do not pose systemic risks are subject to limited requirements, while higher-risk GPAI models must comply with stricter regulations.
Governance Architecture
To ensure effective enforcement of the AI Act, several governing bodies will be established:
- An AI Office within the European Commission to enforce common rules across the EU.
- A Scientific Panel of independent experts to support enforcement activities.
- An AI Board comprising representatives from member states to advise the Commission and ensure consistent application of the Act.
- An Advisory Forum for stakeholders to provide technical expertise.
Penalties for Non-Compliance
The AI Act establishes fines for violations, calculated as a percentage of a company’s global turnover from the previous year or as a fixed amount, whichever is higher. Small and medium enterprises (SMEs) and startups will face proportional penalties.
Support for Innovation
Beyond enforcing existing laws on fundamental rights and safety, the AI Act aims to promote investment and innovation in AI within the EU, fostering the development of a single market for AI applications. This aligns with the EU’s coordinated plan to accelerate AI investment in Europe.
Background and Development
The European Council has been actively discussing AI since October 2020, prompting the European Commission to propose ways to enhance public and private investments in AI research and innovation. The proposal for the AI Act was released in April 2021, aiming to harmonize rules around AI technology.
After considerable discussions, the Council adopted its position on new rules for AI on December 6, 2022, emphasizing the need for safe and trustworthy AI that respects fundamental rights. Following extensive negotiations, a provisional agreement was reached on December 9, 2023, leading to the formal adoption of the Act on May 21, 2024, and its entry into force on August 1, 2024.
As of March 13, 2026, the Council has agreed on additional provisions to prohibit AI practices that involve the generation of non-consensual sexual content or child sexual abuse material, further highlighting the legislation’s commitment to protecting individuals’ rights and safety.
