AI law may cover agent autonomy, deepfakes

📌 Summary:

• MeitY is exploring a standalone AI law (separate from the IT Act, 2000) to regulate artificial intelligence, The Indian Express has learnt
• Likely elements: a consent-based framework for synthetically-generated content (AI-made or altered media, e.g. deepfakes), curbs on agentic AI autonomy, and regulatory sandboxes for high-risk applications
• The IT Ministry's cyber laws division has been asked to audit the IT Act and its rules to find regulatory gaps
• Agentic AI concern: govt will weigh how much autonomy such autonomous, goal-driven agents should have, and whether they can retain and reuse data (agentic AI independently plans multi-step actions, uses tools, and adapts with minimal human supervision)
• Sandbox plan: MeitY to consult RBI and SEBI to set up a regulatory sandbox for AI in high-impact areas like finance and public services
• Liability debate: unlike social media's 'safe harbour' (immunity for user-generated content), AI models editorialise responses that are hard to trace to a source — so safe harbour is hard to translate; govt has asked two private-sector legal experts to submit separate draft liability frameworks

🎯 UPSC Relevance: GS3 — emerging technology governance, AI regulation, data protection; GS2 — regulatory frameworks and platform liability

📝 Prelims Facts:

• MeitY (Ministry of Electronics and IT) is drafting the AI law; likely standalone, not under the IT Act, 2000
• 'Safe harbour' = legal immunity for platforms hosting user-generated content
• Regulatory sandbox to be built in consultation with RBI and SEBI

🔑 Key Term: Agentic AI — autonomous AI systems that, given a high-level goal, independently plan multi-step actions and use tools with minimal human supervision