Context:
The European Union (EU) is poised to enact the world’s inaugural comprehensive legislation for the regulation of Artificial Intelligence (AI). The proposed framework is anticipated to face a parliamentary vote in early 2024, with potential enforcement as early as 2025.
Relevance:
GS II: International Relations
Dimensions of the Article:
- Components of the EU Framework for AI Regulation
- Global Approaches to AI Regulation
Components of the EU Framework for AI Regulation:
Legislative Safeguards:
- Consumer Empowerment: Individuals enabled to file complaints for perceived AI violations.
- Limits on Law Enforcement Adoption: Clearly defined boundaries on AI usage by law enforcement.
- Stringent AI Restrictions: Strong limitations on facial recognition and AI behavior manipulation.
- Penalties for Violations: Provision for severe penalties for companies breaching AI rules.
- Restricted Biometric Surveillance: Government-permitted real-time biometric surveillance in public areas only for serious threats like terrorism.
AI Application Categorization:
- Four Risk Classes: AI applications categorized into four risk levels based on risk and invasiveness.
- Prohibited Applications: Mass-scale facial recognition and behavioral control AI mostly banned, with law enforcement exemptions.
- High-Risk Applications: Allowed with certification and transparency, e.g., AI tools for self-driving cars.
- Medium-Risk Applications: Deployable without restrictions, like generative AI chatbots, with user disclosure, transparency obligations, and detailed documentation.
Other EU Regulatory Milestones:
- GDPR Implementation: Enforced since May 2018, focusing on privacy and explicit consent for data processing.
- Sub-Legislations: DSA and DMA:
- Digital Services Act (DSA): Regulates hate speech, counterfeit goods, etc.
- Digital Markets Act (DMA): Identifies “dominant gatekeeper” platforms, addressing non-competitive practices and dominance abuse.
Global Approaches to AI Regulation:
European Union (EU):
- Stringent Categorization: Classifies AI based on invasiveness and risk levels.
United Kingdom (UK):
- ‘Light-Touch’ Approach: Fosters innovation in AI with a less restrictive regulatory stance.
United States:
- Balanced Position: Positioned between supporting innovation and implementing some regulatory measures.
China:
- Sovereign Measures: Introduces its own AI regulations aligning with national policies and priorities.
India’s Strategy on AI Regulation:
Stance Evolution:
- From Non-Consideration to Formulation: India shifts from not considering AI regulation to actively formulating regulations based on risk and user-harm approach.
Advocacy for Responsible AI:
- Inclusive National AI Strategy: #AIFORALL (2018) focused on inclusivity in AI adoption.
- Principles of Responsible AI (2021): NITI Aayog introduces principles emphasizing equality, safety, inclusivity, transparency, accountability, privacy, and positive human value.
- Comprehensive National Initiative: IndiaAI (March 2023) introduced as a comprehensive program covering all AI-related research and innovations.
- Statutory Authority Recommendation: Telecom Regulatory Authority of India (July 2023) proposes a domestic statutory authority for AI regulation with a risk-based framework and an advisory body with diverse expertise.
Sector-Specific AI Frameworks in India:
Healthcare Sector:
- Ethical Guidelines: Indian Council of Medical Research issues guidelines for AI in biomedical research and healthcare (June 2023).
Capital Market:
- SEBI Circular (January 2019): Guides policies and establishes an inventory for AI systems in the capital market.
Education Sector:
- NEP 2020 Recommendation: National Education Policy recommends integrating AI awareness into school courses.
-Source: Indian Express
