EU AI ACT ENFORCED OFFICIALLY

Authored by Tanima Bhatia

Key Highlights

  1. Comprehensive Regulation Framework: The EU AI Act establishes strict regulations for AI systems, categorizing them into prohibited, high-risk, and general-purpose models to ensure safety, transparency, and respect for fundamental rights.
  2. Implementation Timeline: The Act will enter into force on August 1, 2024, with phased enforcement for different AI systems:
    1. Prohibited systems general Provisions: February 1, 2025
    2. General-purpose AI models and Notifying Authorities: August 1, 2025
    3. Remaining Act: August 1, 2026
    4. High-risk systems (Article 6(1)): August 1, 2027
  3. Implications for the AI Industry: The Act will lead to compliance costs and require significant adjustments by AI developers. It sets a global regulatory precedent and emphasizes ethical AI practices, potentially influencing international AI standards and market dynamics.

Introduction

The European Union Act, officially came into force today i.e. 1st August 2024, officially published in the EU Official Journal on 12th July 2024. This landmark regulation, formally known as Regulation (EU) 2024/1689, marks a significant step in the European Union’s efforts to regulate artificial intelligence (AI) technologies. In this blog, we will explore the key aspects of the EU AI Act, its timeline, and its implications for the AI industry.

Key Provisions of the EU AI Act

The EU AI Act establishes a comprehensive legal framework for AI technologies, aiming to ensure they are safe, transparent, and respect fundamental rights. Here are the main provisions:

  1. Prohibited AI Systems: Certain AI systems are outright prohibited under the Act due to their potential to harm individuals or society. These include AI systems that manipulate human behavior or exploit vulnerabilities in a harmful manner.
  2. High-Risk AI Systems: AI systems deemed high-risk must comply with strict requirements, including risk management, data governance, transparency, and human oversight. These systems are used in critical areas such as healthcare, transportation, and law enforcement.
  3. General Purpose AI Models: The Act sets rules for general-purpose AI models that can be adapted for various applications. These models must meet requirements for accuracy, robustness, and cybersecurity.
  4. Transparency Obligations: AI systems interacting with humans, such as chatbots, must be designed to inform users that they are interacting with AI. This provision aims to ensure transparency and build trust in AI technologies.

Timeline of the Act

The journey began on April 21, 2021, when the European Commission proposed the regulation. Following this, a public consultation period concluded on June 30, 2021, with 304 submissions received. The first compromise text was shared by the EU Council on December 6, 2021, and the Committee on Legal Affairs published its amendments on March 2, 2022. The Council of the EU adopted its general approach on June 1, 2022, leading to the European Parliament’s negotiating position, which was adopted on December 6, 2022, with significant support. A provisional agreement between the Parliament and the Council was reached on June 14, 2023, with all 27 member states endorsing it. By December 9, 2023, the AI Act was approved by the Internal Market and Civil Liberties Committees, followed by the official launch of the European Artificial Intelligence Office. Finally, on February 13, 2024, the European Council formally adopted the EU AI Act, marking a significant step in regulating AI technologies in the EU​. Check our info sheet here.

With the publication of the EU AI Act, a specific timeline for its implementation has been triggered:

  • Entry into Force: The Act enters into force 20 days after its publication, making it effective from August 1, 2024.
  • General Provisions (Chapter 1) and Prohibited Systems (Chapter 2): Six months from the entry into force, the rules regarding prohibited systems and general provisions will be enforced, starting February 1, 2025.
  • Notifying Authorities (Chapter III Section 4), General Purpose AI Models (Chapter V), Governance (Chapter VII), Confidentiality and Penalties (Chapter XII), and Confidentiality (Article 78), Except Fines for GPAI Providers (Article 101): Twelve months from the entry into force, rules for general-purpose AI models, Notifying Authorities, Governance, Confidentiality and penalties, and confidentiality will come into effect, starting August 1, 2025.
  • Application of EU AI Act for AI Systems including Annexure III: Twenty-four months from the entry into force, regulations for high-risk systems under Annex III will be enforced, starting August 1, 2026.
  • Other Risk Systems: Thirty-six months from the entry into force, regulations for other risk systems will be enforced, starting August 1, 2027.

Implications for the AI Industry

The EU AI Act is a significant regulatory step with far-reaching implications for the AI industry. Here are some of the key impacts:

  1. Compliance Costs: AI developers and businesses will incur costs to ensure compliance with the Act’s requirements. This includes investing in risk management systems, transparency measures, and obtaining certifications for high-risk AI systems.
  2. Innovation and Development: While the Act aims to foster innovation, there is concern that stringent regulations could stifle creativity and slow down the development of new AI technologies. However, the emphasis on safety and ethical considerations is expected to enhance public trust in AI.
  3. Global Influence: The EU AI Act sets a precedent for AI regulation globally. Other regions and countries may adopt similar frameworks, influencing international standards for AI technologies.
  4. Market Dynamics: Companies that can swiftly adapt to the new regulations may gain a competitive advantage. Compliance with the EU AI Act could become a selling point, demonstrating a commitment to ethical AI practices.

Penalties under the Act

The AI Act imposes substantial fines for non-compliance, with penalties reaching up to EUR 35 million or 7 percent of global annual turnover for using prohibited AI systems. Other violations can result in fines up to EUR 15 million or 3 percent of worldwide turnover. Providing incorrect, incomplete, or misleading information to authorities can lead to fines up to EUR 7.5 million or 1 percent of annual turnover. For SMEs, including start-ups, the fines are capped at the lower of these percentages or amounts. Providers of general-purpose AI models face fines up to 3 percent of their global turnover or EUR 15 million, whichever is higher.

 

Conclusion

The EU AI Act represents a monumental step in the regulation of artificial intelligence, aiming to balance innovation with safety and ethical considerations. While the implementation of this Act will require significant effort and investment from the AI industry, it also paves the way for a more transparent and trustworthy AI ecosystem. As the timeline unfolds, stakeholders must stay informed and proactive in meeting the Act’s requirements, ensuring that AI technologies continue to benefit society while safeguarding fundamental rights.

Further Reading

For a detailed reading of the Act, you can visit the official journal or find the EU AI Act.

References:

  1. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=OJ%3AL_202401689#d1e2090-1-1
  2. https://www.linkedin.com/posts/luizajarovsky_ai-aigovernance-aiact-activity-7224603875402485760-kNUW/?utm_source=share&utm_medium=member_android
  3. https://www.europarl.europa.eu/doceo/document/TA-9-2024-0138_EN.pdf