As artificial intelligence rapidly integrates into our daily lives, from work to entertainment, the Indian government is moving to ensure its development is safe and responsible. A new framework of safeguards is now under review, signalling a pivotal
moment for tech policy.
What Safeguards Are Being Discussed?
The government is considering a multi-layered approach to AI safety, with a particular focus on powerful, cutting-edge systems often called 'frontier models'. Among the key proposals being examined are mandatory pre-release evaluations to assess a model's capabilities and risks before it reaches the public. This could involve testing in controlled environments, or 'sandboxes', to study an AI's behaviour and patch vulnerabilities. Officials are also looking into security assessments proportionate to the risks posed by an individual AI system. The goal is to establish clear responsibility for any harm caused by these advanced models and ensure human oversight remains a central pillar of their operation. This represents a potential shift from earlier voluntary guidelines to more formal, legally binding obligations.
The Push for a Safety-First Approach
This regulatory push is driven by growing concerns over the potential misuse of AI. Recent warnings from the Indian Computer Emergency Response Team (CERT-In) have highlighted the increasing sophistication of AI-driven cyberattacks. In advisories, CERT-In noted that generative AI is lowering the barrier for cybercriminals to discover software vulnerabilities and craft convincing phishing campaigns. There have also been reports of AI agents accessing confidential data and live systems without authorisation. Beyond cybersecurity, the government is also focused on tackling societal harms like deepfakes, financial fraud, impersonation, and the spread of misinformation, which have become priorities for parliamentary review.
Balancing Innovation and Regulation
While safety is a priority, policymakers are navigating a delicate balance to avoid stifling India's burgeoning AI ecosystem. The government's stated approach has been to foster innovation while managing risks. Officials have acknowledged that excessively broad or rigid rules could place a heavy compliance burden on startups and research institutions, potentially hindering progress. The current framework relies on a mix of existing laws—like the IT Act and the Digital Personal Data Protection Act—and targeted advisories. For instance, a revised advisory from March 2024 moved away from requiring government pre-approval for all models and instead mandated self-certification, bias audits, and clear labelling of AI-generated content. This suggests a preference for a risk-based approach, where the most powerful models receive the highest level of scrutiny, allowing less risky applications more freedom to innovate.
Potential Impact on India's Tech Scene
The proposed safeguards will have significant implications for every company developing or deploying AI in India. For developers of frontier models, a pre_release testing regime could introduce new development costs and timelines. However, it could also build greater public trust in their products. For smaller startups, the emphasis on a risk_based framework may come as a relief, as it would likely spare them from the most stringent requirements applied to high_risk systems. The focus on self_certification and creating an internal 'AI Compliance Register' means companies will need to be diligent about documenting their safety testing, bias audits, and labelling mechanisms. Furthermore, the government's IndiaAI Mission, which includes providing subsidised computing power, aims to support the domestic tech industry through this transition.
How India's Stance Compares Globally
India's approach appears to be charting a middle path compared to other major global powers. The European Union has adopted the comprehensive, legally binding EU AI Act, which categorises AI systems by risk and imposes strict rules on high_risk applications. In contrast, the United States has largely relied on executive orders and voluntary commitments from tech companies while specific legislation is debated. India's model seems to be a hybrid, combining existing laws with new, targeted rules and a strong emphasis on self_regulation and labelling. By preparing a dedicated legal framework for frontier models, India is signalling that while it supports innovation, it will not hesitate to enforce specific rules for systems deemed capable of causing significant harm. This positions India as a key player shaping global norms around responsible AI.
