The Search For AI Guardrails
For nearly 25 years, India’s digital economy has largely been governed by the Information Technology Act, 2000, which was drafted long before artificial intelligence (AI) became embedded in our everyday lives.
But as AI rapidly moves from a niche technology to critical infrastructure, powering everything from customer service and content creation to healthcare, finance and governance, policymakers are increasingly questioning whether the existing legal frameworks are enough.
The debate has gained urgency, as, with the rise of generative AI, countries across the globe are fighting concerns around growing misinformation, deepfakes, algorithmic bias, privacy and accountability.
Amid this, India, too, is grappling with something very perplexing: how to regulate technology that is evolving faster than traditional policymaking?
Recently, Union IT minister Ashwini Vaishnaw said the government believes a new law may be required because “the world of AI is very different from the world when the IT Act was enacted”. Now, as India weighs the need for a dedicated AI law, founders, policy experts and legal practitioners argue that India should adopt a risk-based approach focused on accountability, transparency and liability rather than copying sweeping frameworks like the EU’s AI Act.
But What Do Founders Want From India’s AI Rulebook?
If India drafts an AI law, founders appear less interested in regulating algorithms but in defining rights, responsibilities and accountability. According to Shayak Mazumder, the cofounder and CEO of Adya.ai, which enables enterprise AI infrastructure, any future AI law should start with a fundamental principle: access.
As AI increasingly becomes a layer underpinning education, work and governance, Mazumder argues that citizens should have the “right to AI” much like access to education or healthcare. He also argued that compliance requirements should be tied to the real-world risks and impact of AI systems, with stricter rules for high-risk applications and lighter obligations for lower-risk use cases.
Deepak Subramanian, the founder of YourTribe, an AI-driven recruitment and talent platform, also said that regulation should focus on the impact of AI systems rather than treating every application equally.
“India should regulate outcomes and impact rather than the underlying technology itself,” he said. Moreover, higher-risk deployments would require disclosure, audit trails and mechanisms for human review, while low-risk tools would face lighter obligations.
The Governance Challenge
Many policy experts and legal practitioners believe India should avoid copying broad AI laws such as the EU’s AI Act. Instead, regulation should focus on the risks and real-world impact of AI systems, building on existing laws where possible.
According to them, future rules should prioritise accountability, transparency and risk-based compliance, while recognising that AI systems are dynamic and cannot be governed like traditional software.
Raj Shekhar, a consultant at the Centre for Law, Policy and Governance, NFPRC Foundation, a public policy and research organisation, believes India does not necessarily need a standalone AI Act modelled on Europe’s approach. Instead, policymakers should build on existing frameworks such as the DPDP Act, the IT Act, and sector-specific regulations from institutions like the RBI and SEBI.
“The temptation may be to copy the EU and regulate the technology in toto, but that’s the wrong move. Better to regulate the associated risks and match obligations to the harm a system can cause,” he said.
Mishi Choudhary, the founder of SFLC.in, a legal services organisation that works exclusively on technology, law, and policy, also cautions against adopting a sweeping EU-style framework.
She believes India should focus on public-sector AI accountability, transparency requirements for high-impact systems, safeguards around election-related synthetic media and targeted obligations for genuinely high-risk deployments.
“If India were to adopt an AI law, it should be narrow and targeted. It should be accompanied by procurement reforms and an open AI policy,” Choudhary said. She argued that any future framework should be accompanied by procurement reforms and an open AI policy.
As governments increasingly deploy AI across public services, overreliance on proprietary models and foreign-controlled APIs could create strategic dependencies, particularly at a time when access to frontier AI capabilities is increasingly influenced by geopolitical considerations.
This distinction is becoming increasingly important as AI finds its way into everything from customer support and content generation to lending decisions and healthcare recommendations.
A chatbot writing marketing copy and an AI model deciding whether someone qualifies for a loan cannot reasonably be subjected to the same level of scrutiny.
Instead, experts point to three areas where a future AI framework should focus:
- Liability: When an AI system causes harm, accountability must be clearly defined across model developers, deployers and platforms.
- Risk-Tiering: A chatbot generating marketing copy cannot be regulated the same way as an AI system used in healthcare, lending, hiring or public services. Compliance requirements should scale with risk.
- Transparency: Users should know when AI is involved in decisions that affect them and have a mechanism to challenge those outcomes.
The Regulatory Fine Line
Even among supporters of regulation, there is widespread concern that getting the framework wrong could create unintended consequences.
India currently sits at a unique moment in the AI race. It possesses one of the world’s largest developer communities, a growing AI startup ecosystem and a massive domestic market that can serve as a testing ground for AI products. Heavy compliance requirements could threaten that advantage.
Unlike large technology firms, startups often lack dedicated legal, policy and compliance teams. Requirements such as extensive audits, certification regimes or complex reporting obligations could disproportionately burden younger companies with limited resources.
This risk has become particularly relevant as AI infrastructure remains concentrated globally. Many Indian startups still rely on foreign frontier models to build products and services. If future regulations require extensive sovereign AI capabilities before local alternatives become globally competitive, some businesses could find their product roadmaps disrupted overnight.
Experts, therefore, argue that compliance obligations should be proportional to company size, revenue and deployment scale. Choudhary, for instance, advocates for regulatory sandboxes and safe harbours for startups, noting that smaller companies cannot be expected to navigate the same compliance burden as large technology firms.
For now, the challenge for policymakers is balancing two competing realities. Move too slowly, and harmful applications may outpace safeguards. Move too aggressively, and innovation may consolidate among a handful of large players capable of absorbing regulatory costs. For a country hoping to become an AI powerhouse, that balance could determine whether regulation becomes a catalyst for innovation or a brake on it.
Top Stories From India & Around The World
- Sarvam Joins Unicorn Club: Bengaluru-based AI startup Sarvam has become India’s 130th unicorn after raising $234 Mn in a $300 Mn Series B round led by HCLTech at a $1.5 Bn valuation. The startup plans to use the capital to build next-generation AI models focused on agentic AI, coding, and cybersecurity while expanding its compute infrastructure and enterprise deployments.
- Avataar’s Low-Cost AI Video Bet: Avataar has launched Varya, an India-built AI video generation model that claims to create videos at just ₹0.50 per second. Built on Alibaba’s open-source Wan 2.2 architecture, the startup says Varya delivers up to 27X faster generation while aiming to make AI video affordable for businesses, creators, and educational institutions.
- Anthropic Suspends Access To Fable5: Anthropic temporarily suspended access to its frontier AI models Fable 5 and Mythos 5 after receiving a US government export-control directive. The move has reignited discussions around India’s dependence on foreign AI platforms, with industry leaders, including Sridhar Vembu, Hemant Mohapatra and Mohandas Pai, calling for stronger sovereign AI capabilities.
- Equal AI Raises $30 Mn: Hyderabad-based Equal AI has raised $30 Mn in a Series B round led by existing investors, including Prosus Ventures and Tomales Bay Capital. The startup plans to expand beyond AI-powered call management into shopping, financial services, communications, and lifestyle use cases as it seeks to become a broader consumer AI assistant platform.
- Zoho Builds Its Own AI Servers: Zoho has unveiled Nathu La, an in-house server platform designed to lower AI inference costs and reduce dependence on foreign technology infrastructure. The company claims the platform can cut total ownership costs by 20-30% and reduce power consumption by up to 18%, helping make AI deployments more cost-efficient.
The Weekly Buzz: Sridhar Vembu Takes A Dig At Salesforce
A debate around enterprise software pricing gained traction last week after Zoho founder Sridhar Vembu amplified concerns about Salesforce’s pricing model.
The discussion centred on a familiar complaint in enterprise software: attractive discounts during the first year followed by significantly higher renewal costs. Vembu warned businesses to be cautious of such pricing structures, arguing that customers often find themselves locked into platforms after making substantial investments in integrations, workflows, and employee training.
The remarks struck a chord with founders, operators and software buyers, many of whom shared experiences around enterprise software pricing, contract negotiations and vendor lock-in. Others criticised the complexity, cost and usability of large enterprise software platforms, arguing that switching becomes increasingly difficult once organisations are deeply embedded within an ecosystem.
The conversation reflects a broader shift underway in enterprise technology. As AI lowers the barriers to building software and intensifies competition, customers are increasingly scrutinising not just product features but also pricing transparency, long-term value and vendor dependence.
The larger takeaway is that in the AI era, enterprise software companies may face growing pressure to justify not only their technology advantages but also the economics of staying on their platforms.
Startup In The Spotlight: JupiterBrains
As enterprises increasingly adopt AI across business workflows, many are discovering that larger frontier models often entail higher inference costs, governance challenges, and deployment complexity. Hyderabad-based JupiterBrains is betting that smaller, domain-specific AI models can offer a more practical path to enterprise AI adoption.
Founded in 2025 by Nilesh Potdar, a former executive at Microsoft, Amazon and Goldman Sachs, JupiterBrains is building a small-model-first AI infrastructure platform focused on enterprise use cases such as BFSI, compliance, onboarding and document intelligence.
The startup’s flagship product, JupiterBrains AI Workbench, combines model routing, governance, observability, deployment tooling and workflow orchestration into a unified enterprise AI stack. The platform is designed to help organisations deploy and manage AI applications while maintaining control over performance, security and compliance requirements.
JupiterBrains focuses on areas such as enterprise document intelligence, KYC and onboarding automation, AI-powered workflow orchestration, compliance management, auditability systems, and multilingual enterprise AI deployments.
By leveraging smaller, specialised AI models, the startup aims to reduce inference costs while providing enterprises with greater privacy, governance and deployment flexibility across cloud, hybrid and on-premises environments.
Operating in India’s growing enterprise agentic AI market, which is projected to reach $1.8 Bn by 2030, JupiterBrains claims it has already generated about ₹70 Lakh in consultancy revenue while expanding its recurring subscription business.
Prompt Of The Week
What prompts and hacks are CTOs, CEOs and cofounders using these days to streamline their work?
Here’s the prompt used by Sajeev Viswanathan, founder and CEO of MiFiX.ai, to assess where AI can deliver the highest business impact and help management articulate a credible AI strategy to shareholders.
“You are the Managing Director of ABC Ltd.
At your company’s AGM, you are asked to elaborate on your AI plans for the company in the immediate and long-term future by a significant shareholder.
Looking at the company’s businesses and financial statements, identify the top 5 opportunities for AI intervention that will enhance shareholder value. Once done, do the following:
- Draft a response highlighting specific areas where you are focusing on
- Include the potential financial and non-financial benefits.
- Include any successful projects
- Partners you are currently working with
- And the cost outlay.
You may also evaluate any statements made by the company’s management team to develop your response.”
Editor’s Note: Some prompts may need to be adjusted by users for best results or may not work as intended for certain users.
[Edited by Shishir Parasher]
[Creatives by Abhyam Gusai]
The post The Search For AI Guardrails appeared first on Inc42 Media.
No comments