‘There is still a dearth of more nuanced AI-specific legislation to govern this evolving sphere’
| Photo Credit: Getty Images
The allocation of ₹2,000 crore to the IndiaAI mission in the Union Budget 2025-26 demonstrates India’s commitment to developing its artificial intelligence (AI) infrastructure. But this also calls for more robust AI regulations. India has legislation governing personal data protection and the government has issued warnings against AI-generated content. However, there is still a dearth of more nuanced AI-specific legislation to govern this evolving sphere. The United States and the state of its AI legislation and regulations serve as a cautionary tale against having an inadequate regulatory approach towards a technology that is effectively uncharted territory.
What is happening in the U.S.
Instead of a comprehensive law that governs AI, the U.S. regulation is fragmented, and the orthogonal views of the Biden and Trump administrations further complicate the matter. For instance, the Biden administration signed an executive order in 2023 on ‘Safe, Secure and Trustworthy Artificial Intelligence’ that broadly established safety, security and transparency guidelines. However, the Trump administration rescinded many of these directives in a new executive order signed on January 23, 2025. Moreover, in the absence of any federal legislation, U.S. States have begun to develop their own regional legislation, examples being New York and Tennessee laws. Although in some cases this can be beneficial, considering different regional contexts, in this case it leads to inconsistency that can potentially make it difficult to navigate legal issues for individuals and tech companies alike.
The patchwork that is the U.S.’s AI regulation landscape can be seen in recent developments. For instance, Texas recently banned foreign AI applications such as DeepSeek on government devices in light of security concerns. However, the app is still at the top of the app store and is widely available elsewhere. On another note, the many ongoing legal battles, such as the lawsuits against OpenAI over training data, exemplify how current legislations on intellectual property need to be updated to be applicable to AI.
India’s focus and what needs to be done
Circling back to India’s budgetary allocations, it is worth noting that almost 44% of the Budget’s allocation towards AI will be directed towards building domestic computing capabilities. According to the Union Finance Minister’s Budget speech, a new centre of excellence for artificial intelligence (AI) for education with an outlay of ₹500 crore will be set up. This follows an announcement in 2024 that AI centres of excellence for agriculture, health care and sustainable cities would be set up. To build India’s capabilities and keep up with emerging industries, the Ministry of Electronics and Information Technology (MeitY) is also setting up five national centres for skilling in the fields of electronics and information technology. The Budget also focuses on developing several strategic components such as compute infrastructure, datasets, the promotion of AI-based applications and innovation centres that will support AI research. Union Minister for IT Ashwini Vaishnaw also announced India’s intention to build its own large language model (LLM) days after DeepSeek R1 was released to the public.
These are all positive developments in India’s bid to become a leading player in the global AI landscape despite falling behind countries such as the U.S., Israel and China. In light of this commitment towards developing its AI ecosystem, India must develop new legislation that cater to the unique problems posed by the technology and modify older legislation to include these requirements.
Arguably, the first step that the Indian government needs to take is establishing a dedicated independent regulatory authority on AI, with representatives from different Ministries and sector-specific bodies to ensure harmonised policies. Once inputs have been taken from various regional and ministerial representatives, clear ethical guidelines and accountability standards that address transparency, algorithmic fairness and human oversight need to be implemented. Regular audits need to occur to ensure these standards are met. With more indigenous AI systems, this will become increasingly important. Collaborating with the Bureau of Indian Standards (BIS) to develop national AI safety and reliability benchmarks would help ensure that AI systems meet these standards.
On IP and copyright
Current intellectual property and copyright frameworks need to be revised to accommodate fair use for AI training without overstepping data privacy laws. Additionally, a licensing framework for copyrighted material needs to be implemented for AI training to ensure that content creators are protected while supporting innovation. Additionally, the Digital Personal Data Protection Act (DPDPA) also needs to be expanded to have specific guidelines on using training data, consent, and anonymisation in AI. Protocols need to be established for text and data-mining to ensure the ethical sourcing of training data.
Specific guidelines and frameworks for accountability and transparency will also have to be established that cater to the use of AI for military purposes, where the margin of error is much lower. The outcry over Israel’s use of AI target-identification systems highlights the need for such frameworks. On this front, the AI Governance Guidelines Development report by MeitY is a significant step in the right direction.
While every new technology is met with doubts and scepticism, from automobiles to the Internet, if history is any indication, it eventually becomes integrated with society despite the risks it may pose. With AI as well, rather than shirk away from it, governments should try and adapt it for their contexts as India is doing. However, this needs to be done responsibly. With technology that is as rapidly evolving as AI, it is integral to change regulations and legislation to meet new and previously unanticipated challenges that it may pose.
Adya Madhavan is a researcher at the Takshashila Institution
Published – February 25, 2025 04:30 pm IST