By - King Stubb & Kasiva on April 9, 2024
On March 13th, 2024, history was made as the European Union (EU) took a decisive step forward in regulating the rapidly advancing field of Artificial Intelligence (AI). The EU Parliament approved the groundbreaking Artificial Intelligence Act with 523 votes in favor, 46 against, and 49 abstentions[1], marking a pivotal moment in shaping the global framework of AI governance. This comprehensive legislation sets out to govern the development and use of AI systems and applications within the EU, ensuring that the technology is safe, transparent, and aligned with fundamental rights.
At the core of the Act lies the commitment to safeguard fundamental rights, democracy, and the rule of law. It introduces measures to mitigate the risks associated with high-risk AI applications, such as biometric identification systems and manipulative AI practices. By prioritizing the protection of citizens’ rights, the EU aims to build trust in AI technologies and promote ethical AI practices.
One of the central components of the Act is the imposition of bans on certain AI applications deemed to pose significant risks to individuals and society. These include:
By imposing strict regulations on these high-risk AI applications, the EU seeks to promote the responsible use of AI across various sectors.
High-risk AI systems are subject to stringent guidelines to ensure clarity, accuracy, and accountability. Developers are now obligated to meticulously assess and mitigate potential risks, maintain detailed usage logs, and establish mechanisms for human oversight. Additionally, consumers possess the right to lodge complaints and receive comprehensive explanations regarding decisions made by high-risk AI systems impacting their rights. This framework promotes confidence and trust in AI deployment while safeguarding individual rights and interests.
Turning to the regulatory aspects, it is vital that General-purpose AI (GPAI) systems meet specific requirements. These encompass not only compliance with EU copyright regulations but also the mandate to furnish detailed summaries of the training data used. More robust GPAI models undergo even stricter scrutiny to identify and address potential systematic risks. Furthermore, the legislation emphasizes the importance of clear labeling for artificial or manipulated media, such as deepfakes.
The EU’s AI Act has garnered mixed reactions from industry stakeholders and policymakers worldwide. While some celebrate its proactive approach, others raise concerns about its potential impact on innovation and competitiveness. Tech giants, startups, and industry associations are closely monitoring the implementation of the Act, anticipating its effects on market dynamics and regulatory compliance. Moreover, the EU’s regulatory framework is likely to influence other countries’ approaches to AI governance, paving the way for international collaboration and standard-setting.
Effective enforcement mechanisms are essential to ensure compliance with the AI Act’s provisions. National authorities and EU agencies play an important role in monitoring and enforcing regulatory compliance, imposing penalties for violations and promoting adherence to ethical standards. Collaboration between policymakers, industry stakeholders, civil society organizations, and academic institutions is key to shaping AI regulation and promoting responsible AI development.
Education and awareness-raising initiatives are necessary to empower stakeholders with the knowledge and skills needed to navigate the complexities of AI regulation. Initiatives promoting digital literacy and responsible AI use are essential for informed decision-making and ethical AI deployment. As AI continues to evolve, policymakers must anticipate future trends and emerging technologies that may pose challenges or opportunities for regulation. By staying abreast of technological advancements, regulators can adapt regulatory frameworks to address evolving risks and promote innovation in the AI ecosystem.
| PROVISION | DESCRIPTION |
| Protection of Fundamental Rights | Introduces measures aimed at safeguarding fundamental rights, democracy, and the rule of law amidst the presence of high-risk AI applications. |
| Restrictions on High-Risk AI | Imposes bans on certain AI applications deemed to pose significant risks to individuals and society, such as biometric identification systems and manipulative AI practices. |
| Transparency and Accountability | Requires high-risk AI systems to adhere to clear obligations aimed at ensuring transparency, accuracy, and accountability. Consumers have the right to lodge complaints and receive explanations regarding decisions made by AI systems affecting their rights. |
| Ad-Hoc AI Office | Establishes an ad-hoc AI office within the European Commission to supervise general-purpose AI systems integrated into other high-risk systems. Supported by an advisory forum comprising representatives from various sectors. |
The Act is currently undergoing a final review by legal experts and linguists, with expectations for official legal status to be achieved around May or June 2024. This process, known as the corrigendum procedure, ensures that any necessary adjustments or clarifications are made before the law is officially enacted. Implementation of its provisions will proceed gradually, with countries mandated to prohibit AI systems deemed harmful six months after the regulations are enacted.
Regulations concerning general-purpose AI systems, such as chatbots, will take effect one year following the enactment of the law. By mid-2026, the comprehensive set of rules, encompassing requirements for high-risk systems, will be fully operational. Breaches of the AI Act may result in fines of up to 35 million euros ($38 million), or 7% of a company’s global revenue.[2]
Upon publication in the official Journal, the regulation will come into effect twenty days later. However, its full applicability is set for 24 months following its enactment. Certain provisions, such as bans on prohibited practices, will take effect six months after the regulation's entry into force. Additionally, codes of practice will be implemented nine months post-enactment, while general-purpose AI rules and governance measures are scheduled for enforcement twelve months thereafter. High-risk systems will have a longer implementation period, with obligations becoming applicable 36 months after the regulation takes effect.
The EU’s AI Act represents a significant milestone in shaping the future of AI governance. By prioritizing the protection of fundamental rights and promoting responsible AI practices, the EU has demonstrated its commitment to harnessing the benefits of AI while mitigating potential risks. The comprehensive framework outlined in the AI Act sets a global standard for AI regulation and paves the way for a future where AI is harnessed responsibly to foster innovation, ensure safety, and uphold fundamental values.
[1] https://www.europarl.europa.eu/news/en/press-room/20240308IPR19015/meps-adopt-landmark-law.
[2] https://www.europarl.europa.eu/topics/en/article/20230601STO93804/.
King Stubb & Kasiva,
Advocates & Attorneys
Click Here to Get in Touch
New Delhi | Mumbai | Bangalore | Chennai | Hyderabad | Mangalore | Pune | Kochi | Kolkata
Tel: +91 11 41032969 | Email: info@ksandk.com
