AI law may cover agent autonomy, deepfakes, Pg17

Key Highlights:

• The Ministry of Electronics and Information Technology (MeitY) is considering a dedicated law for Artificial Intelligence (AI) in India.
• The proposed law aims to address synthetically-generated content through a consent-based framework and regulate the autonomy of agentic AI.
• Two legal experts have been tasked with drafting separate liability frameworks for AI models.
• The government previously held that existing laws were sufficient but is now rethinking its approach due to the rise of deepfakes and concerns over agentic AI.

Detailed Insights:

• The new AI law is expected to be a standalone piece of legislation, separate from the Information Technology Act, 2000.
• MeitY is analyzing the IT Act and its rules to identify existing regulatory gaps concerning AI.
• The government plans to consult with the Reserve Bank of India (RBI) and the Securities and Exchange Board of India (SEBI) to establish regulatory sandboxes for high-risk AI applications, especially in finance.
• A key focus is determining the liability of AI platforms and companies for content generated by their models, contrasting with the safe harbour provisions for social media.
• The Digital Personal Data Protection Act, 2023 addresses broader privacy issues, but the AI law will specifically examine consent and data ownership in the context of AI.
• India's shift towards a comprehensive AI law follows increasing instances of objectionable deepfakes and concerns about autonomous AI systems.
• The government has previously mandated social media companies to label AI-generated content.

Key Concepts Involved:

• Synthetically-generated content: Digital media created or altered by AI, rather than humans.
• Agentic AI: Autonomous AI systems that plan and execute multi-step actions with minimal human supervision to achieve a high-level goal.
• Regulatory Sandbox: A controlled environment where new financial products, services, or business models can be tested without immediately incurring all the normal regulatory consequences.
• Safe Harbour: Legal immunity granted to online platforms from liability for user-generated content they host.
• Deepfakes: Synthetic media in which a person in an existing image or video is replaced with someone else's likeness using AI.