In a push towards its own military artificial intelligence, the Defence Research and Development Organisation has started working towards an AI solution designed specifically for cyber warfare, vulnerability discovery, malware analysis, and threat intelligence. The project aims to create a defence-grade AI capability that can operate entirely within secure military networks, reducing dependence on foreign AI models for sensitive national security applications. As per officials, the system will enhance the cybersecurity of defence networks by helping identify weaknesses before they can be exploited by adversaries.
Officials say that the intended system will function in a fully air-gapped environment, and it will be ensured that all model weights,
training data, inference logs, and operational outputs remain within defence-controlled infrastructure.
The initiative comes at a time when the Indian defence establishment is increasingly focusing on artificial intelligence, cyber warfare, and emerging technologies as components of future military capability. The Ministry of Defence declared 2025 as the “Year of Reforms” with emphasis on technology-driven transformation, including artificial intelligence, machine learning, and cyber capabilities. The system will address what is described as the risks associated with dependence on foreign-developed AI systems for national security applications. This includes concerns related to data, potential backdoors, adversarial model poisoning, and dependence on overseas computing infrastructure. The system is expected to identify a wide range of software vulnerabilities, including memory corruption bugs and use-after-free and double-free vulnerabilities, as well as authentication bypasses, injection attacks, cryptographic weaknesses, and logic flaws.
The model is expected not only to detect vulnerabilities but also to generate proof-of-concept exploit code to help analysts validate security flaws. It will additionally recommend context-aware patches and explain remediation measures for identified vulnerabilities.
In terms of scale, DRDO envisions an LLM model in the 30-70 billion parameter category, placing it in the class of advanced large language models currently available globally. The system is expected to incorporate technologies such as Retrieval-Augmented Generation (RAG), agentic reasoning frameworks, and human-in-the-loop reinforcement mechanisms to continuously improve performance.
Once developed, the capability is expected to be deployed across cyber laboratories, service headquarters, and designated cyber defence units.
The initiative comes months after senior DRDO officials publicly stressed that India cannot rely on foreign AI models for critical military applications and highlighted the need for trusted indigenous AI systems for defence and national security use.
If realised, the project could become one of India’s first dedicated military-grade AI platforms focused specifically on cyber defence, vulnerability discovery, malware analysis, and threat intelligence operations.
/images/ppid_a911dc6a-image-178134322263152260.webp)
