India Needs To Build AI Defence Capabilities As Models Get More Powerful
Last Updated: 2026-09-26 12:53:10
As artificial intelligence models become more capable, building stronger defences around AI systems could become just as important as developing the technology itself. The need for AI security and defence was a key part of the discussion among Indian AI startup founders at the Moneycontrol Startup Conclave 2026.
The issue came up during a panel titled “India’s AI Unicorns: Models, Moats & the Race to Scale” on September 26. The session featured Pratyush Kumar, Co-Founder of Sarvam, and Mukund Jha, Founder of Emergent, and was moderated by Nalin Mehta, Managing Editor, Moneycontrol.
Latest and Popular Mobiles
More Powerful AI Could Also Mean More Vulnerabilities
Jha said AI safety is a serious issue as models continue to become smarter and more capable.
Related Articles
His concern is that increasing capability does not automatically remove the risks associated with AI. As models become more powerful, vulnerabilities in systems can potentially become more consequential.
Jha said AI systems could potentially exploit vulnerabilities, making security and defence an important part of the development of increasingly capable AI.
The discussion therefore went beyond the question of how quickly companies can build better models. It also raised the question of whether businesses and countries are prepared to defend the systems and infrastructure around them as AI capabilities advance.
‘We Need Defence’
For Jha, the answer is not simply to slow the development of AI. He said commercial pressure is strong enough that he does not expect the development of increasingly capable AI systems to stop. Instead, he argued for building the necessary defences alongside the technology.
This creates a different way of looking at the AI race. Progress in AI models and applications needs to be accompanied by progress in cybersecurity and other defensive capabilities that can protect AI systems from potential misuse or exploitation.
Jha’s comments came as Emergent itself is moving towards more autonomous AI systems. The company is expanding from software creation towards autonomous business automation, with AI agents capable of taking on more tasks.
That makes the security question particularly relevant as AI moves from generating responses to carrying out actions within business workflows.
India Should Accelerate AI, Says Sarvam
Pratyush Kumar offered a broader perspective on India’s AI ambitions during the same discussion.
He described the current phase as “year one” for India’s AI journey, with the country now building capabilities around models, GPUs and talent.
Kumar argued that AI should be accelerated from India’s perspective rather than slowed down, while also building the necessary defences.
His argument was linked to the potential scale of AI adoption in India. He said India has advantages in areas such as scale, talent and adoption, and suggested that the country could become the largest user of AI technology in the future.
That means security cannot be treated as a separate issue that comes after AI adoption. As more AI systems are deployed across businesses, government services and other areas, the ability to protect those systems becomes increasingly important.
AI Is Moving Into Real-World Systems
The panel also highlighted how Indian AI companies are moving beyond consumer-facing chatbots and into real-world applications.
Kumar spoke about AI applications in agriculture, education and healthcare. He also discussed projects involving the digitisation of large volumes of documents and the potential to use AI to reach people at population scale.
Sarvam is also building its own infrastructure, including a GPU cluster, software and inference stacks. Kumar said the company is working to host its own models as well as open models and is building capabilities around documents, voice, video, coding and cybersecurity.
As AI becomes embedded in more of these systems, the potential impact of security failures also becomes more significant. The panel’s discussion suggested that building AI capability and building the ability to defend that capability need to happen together.
The AI Race Is Not Just About Bigger Models
The discussion at Moneycontrol Startup Conclave 2026 presented AI progress as a race involving more than model performance.
Companies are working to make AI models smarter, cheaper and more useful, while AI agents are taking on increasingly complex tasks. At the same time, founders are recognising that stronger capabilities bring new security challenges.
Jha’s warning was that AI systems could potentially exploit vulnerabilities as they become more capable. Kumar’s position was that India should continue accelerating AI development while building defences.
For India, that means the AI opportunity discussed at the panel is not only about creating models or deploying applications. It is also about developing the security capabilities required to operate increasingly powerful AI systems safely.