Knowledge Exchange on the EU AI Act and Classification of AI Systems

The growing role of AI in strategic decision-making is undeniable, owing to its ability to analyze vast amounts of data. However, as AI becomes more integrated into human decision-making processes, ethical concerns around privacy, surveillance, bias, and discrimination have emerged. These concerns highlight the need for careful oversight to protect fundamental human rights. The EU AI Act represents a significant step forward in creating a robust regulatory framework for AI technology. In parallel, Indonesia launched its National AI Strategy in 2020. The Knowledge Exchange served as a platform for stakeholders from the EU, Indonesia and Germany to share insights and best practices on AI governance.

EU AI Act and Indonesia´s National AI Strategy

Ms. Elinor Wahal, Legal and Policy Officer of the EU AI Office, European Commission highlighted the long-term process of drafting the EU AI Act, which began in 2018 when the European Strategy on AI, High-Level Expert Group on AI and European AI Alliance coordinated and drafted a plan on AI. The goal was to create a legal framework that safeguards rights and mitigates safety risks associated with AI, while fostering trust and excellence in AI systems. The EU AI Act upholds these goals by classifying AI systems based on risk levels and the data they process, establishing a pioneering framework for AI governance. It outlines a risk-based framework for AI systems, categorizing them into four risk levels: unacceptable, high, limited, and minimal. AI systems deemed to pose an unacceptable risk, are prohibited, while high-risk systems—those impacting critical sectors such as healthcare, transportation, and public safety—are subject to stringent requirements, including risk assessment, transparency, and human oversight. Limited and minimal-risk systems face fewer obligations, primarily focusing on user transparency. This tiered approach aims to balance innovation with public safety and trust in AI. High-risk AI systems are permitted but must meet stringent standards for data quality, transparency, human oversight, accuracy, and cybersecurity. These systems are also subject to third-party conformity assessments and must be registered in a new EU database. To ensure effective implementation, enforcement will be carried out by national competent authorities and the AI Office. This will be supported by close collaboration with Member States and additional technical expertise.

In Germany, the implementation of the EU AI Act is based on four principles. Mr. Jan Bernotat, Policy Officer Artificial Intelligence, German Ministry for Economic Affairs and Climate Action (BMWK) outlined the principles which focus on user centric structures, establishing a one-stop shop for AI regulation, harmonizing regulations and consolidating administrative competencies. These four principles are important to ensure the easy access to determine AI systems conformity towards the Act and ensure no overlapping regulations. These key principles are also part of the German National Strategy on AI.

In Indonesia, the strategy on AI is delineated in the Indonesia National Strategy on AI 2020 – 2045. Mr. Hario Bismo Kuntarto, Team Leader for Digital Economy, Game and Artificial Intelligence, Directorate General of Informatics Applications, MCDA underlined that the strategy aims to build a sustainable and inclusive AI governance ecosystem, ensuring universal access to AI, fostering digital talent and safeguard AI usage. The strategy is aligned with ASEAN´s AI Governance and Ethical Guidelines and UNESCO´s Ethics on AI Usage Guidelines.

AI Developments and Regulatory Challenges

Prof. Dr. Ir. Hammam Riza, Head of Indonesia’s Intelligence Industry Research and Innovation Collaboration (KORIKA) highlighted the progress Indonesia has made in implementing the National AI Strategy. Since its publication in 2020, the country has adopted a holistic approach, organizing numerous workshops and discussions to address issues such as misinformation, security, and the impact of automation on jobs and socio-cultural stability. However, challenges remain, including the need for greater inclusivity, ethics, and ensuring AI accountability and transparency.

Mr. Maximilian Hoeving, Senior Regulatory Expert at Siemens AG, discussed the complexities of harmonizing AI regulations with existing industrial laws. As AI becomes increasingly integrated into industrial applications, the need to balance innovation with regulation becomes critical. The EU AI Act, while offering necessary safeguards, must also account for potential overlaps with pre-existing machinery laws to avoid stifling innovation

Outcomes and Next Steps

The Knowledge Exchange underscored the mutual benefits of international collaboration on AI governance. Both Indonesia and Germany recognized the importance of a balanced approach that fosters innovation while addressing ethical and regulatory challenges. Moving forward, the participants emphasized the need for continuous dialogue, knowledge sharing, and alignment between national AI strategies and international frameworks. The next steps involve deeper collaboration between both countries, focusing on strengthening AI governance frameworks, addressing emerging challenges, and ensuring the responsible development and deployment of AI technologies.