Neural Network: September Edition
A Monthly Newsletter on AI Policy and its Multistakeholder Regulation, from the AI Knowledge Consortium
Feature: AIKC Research
AIKC organised a workshop, Rethinking Liability Regimes in an AI World, this May, to evaluate the suitability of existing liability regimes for users, publishers and intermediaries harnessing AI systems. Members, the think-tanks Esya Centre and Cybersaathi collaborated to author a comprehensive report based on the theme.
“Crafting a Liability Regime for AI Systems in India” attempts to address the questions surrounding the creation of liability rules around AI technologies. Specifically, it tries to tackle the debate on whether AI systems merit the blanket application of strict liability rules or a more contextualised and targeted framework.
Read it here.
Recent Indian Developments in AI Policy
Revisions in requirements for the INDIAai tender
The Ministry of Electronics and Information Technology (MeitY) issued a corrigendum revising the requirements of the INDIAai tender for high-speed computing AI infrastructure
The Ministry of Electronics and Information Technology (MeitY) halved the revenue threshold for companies seeking to bid for the Rs 10,000 crore graphics processing units (GPUs) INDIAai tender, from ₹100 crore to ₹50 crore. A number of other changes have also been announced, including mandating adherence to the Make in India guidelines, and setting requirements for energy efficiency. While MeitY has heeded the suggestions made by smaller AI companies, it remains to be seen whether such players will benefit from the changes
Global AI Governance
California Enacts Regulations on Deepfakes - but not AI
California's landmark laws on deepfakes set the stage for a safer digital landscape
California has enacted five laws regulating deep fakes to safeguard election integrity and performers' rights. Three laws target election-related deep fakes: one immediately bans deceptive content, enforceable 120 days before and 60 days after elections; another requires labels on AI-generated political ad content, effective January 2025; and the "Defending Democracy from Deep Fake Deception Act" mandates social media platforms label or remove AI deep fakes within 72 hours of complaints, also effective January 2025. Additionally, two other laws protect performers' rights by requiring consent for using deep fakes and mandating estate permission for depicting deceased performers in commercial media. These regulations aim to prevent misinformation and ensure a safer digital landscape ahead of the 2024 Presidential election.
Simultaneously, California Governor Gavin Newsom vetoed SB 1047, a bill that proposed to regulate large AI systems by requiring safety testing and allowing the State to hold companies accountable for AI-related harms. Governor Newsom argued that the bill was too narrowly focused on frontier AI models, overlooking risks in other areas.
Council of Europe Adopts First Ever Legally Binding AI Treaty
US, EU, UK, and Japan among key signatories of the Framework Convention on Artificial Intelligence, Human Rights, Democracy, and the Rule of Law
The Council of Europe has adopted the first-ever legally binding AI treaty Framework Convention on Artificial Intelligence, Human Rights, Democracy, and the Rule of Law with support from the US, EU, UK, and Japan. The treaty establishes a global framework to ensure AI systems uphold human rights and democracy. It covers the use of AI systems in the public sector – including companies acting on its behalf – and in the private sector. It mandates transparency, accountability and oversight, requiring governments to implement safeguards and potential bans where AI poses significant risks.
UN Adopts Pact for the Future
The Summit of the Future sets ambitious goals for AI governance, sustainable development, and global cooperation
The UN adopted Pact for the Future, which includes the Global Digital Compact (GDC) and a Declaration on Future Generations outlining comprehensive goals for AI governance, sustainable development, international security, and youth engagement. GDC also draws on the Governing AI for Humanity, a report released this month by the UN’s High-Level Advisory Body on AI (HLAB-AI), which proposes an International Scientific Panel on AI and a global AI capacity network. Next step will be to launch a Global Dialogue on AI Governance to address AI’s impact on human rights and development.
AIKC secretariat’s researchers also co-authored an article for the Print that outlines different approaches countries may take in implementing the GDC. They push for practical and global frameworks (such as technical standards).
Work by AIKC Members:
Responsible AI: Aapti Institute released three papers—All About the Bias: Mapping and Mitigating Bias Across the AI Value Chain, Rewiring Responsible AI: From Principles to Practice, and Decoding AI Development: The Efficacy of the Value Chain Ontology—focused on responsible AI (R-AI) through the lens of the AI value chain, emphasising accountability, risk mitigation, and bias reduction.
#IGPPExpertTalks-AI Special Series: The Institute of Government and Public Policy (IGPP) hosted several expert-led sessions exploring Responsible AI, its societal impact, and how global AI ethics and standards take into account diverse cultural values. Key discussions also included AI and gender, shaping inclusive technology and addressing AI-generated misinformation during elections.
Can AI Light Up India’s Energy Shift? A Pragmatic Look at India’s AI and Energy Transitions: Goa-based Transitions Research’s blog highlights AI's potential in optimising India’s energy transition but stresses that its success depends on addressing governance, infrastructure and data challenges.
Final Thoughts
Just launched: BharatGen, the first government-funded multimodal large language model project to create efficient and inclusive AI in Indian languages. The project will be spearheaded by IIT Bombay under the National Mission on Interdisciplinary Cyber-Physical Systems. It aims to overcome the existing lack of training data in regional languages which stymied their development.
Could such projects herald the expanded use of AI in Tier III and III cities in India?
We’d love to hear your thoughts. Email: Secretariat@aiknowledgeconsortium.com.