India's government has tasked homegrown AI leaders Sarvam AI and BharatGen with developing advanced cybersecurity AI models similar to Anthropic’s Mythos, aiming to safeguard critical infrastructure through sovereign technology solutions hosted on isolated domestic clouds.
- Sarvam AI and BharatGen to develop cybersecurity AI akin to Anthropic’s Mythos
- CERT-In currently tests a hybrid of indigenous and open-source models for critical infrastructure security
- Government seeks on-premises AI deployment to avoid foreign cloud dependency
What happened
The Indian government has commissioned two domestic AI firms, Sarvam AI and BharatGen, to create advanced cybersecurity capabilities modeled after Anthropic's Mythos. These new AI models are planned to operate on isolated government compute infrastructure to protect critical national assets from cyber threats.
Currently, India relies on a blend of indigenous and open-source AI technologies, including outputs from Sarvam AI, to detect security weaknesses. The national cybersecurity agency, CERT-In, is actively testing alternative AI models within a controlled environment to evaluate and improve protections for critical infrastructure.
Why it matters
The growing prevalence of cyberattacks and faster exploitation timelines for unknown software vulnerabilities have heightened the urgency for sovereign AI cybersecurity tools. India recorded a sharp rise in cyber incidents, nearly doubling from 1.4 million to 2.9 million, emphasizing the need for robust defenses that are not dependent on foreign technology providers.
By developing indigenous AI models similar to Mythos, India aims to maintain control over sensitive cybersecurity tools and avoid potential risks associated with overseas cloud deployments. This initiative aligns with broader government priorities to reduce dependence on foreign AI and enhance national digital security resilience.
What to watch next
The timeline for the development and deployment of these Mythos-like AI models remains unspecified. However, attention will focus on how effectively Sarvam AI and BharatGen can meet or exceed the capabilities currently delivered by combined CERT-In AI efforts, which have achieved about 60-70% of Mythos’ functionality in vulnerability detection and remediation.
India’s ongoing negotiations with Anthropic and the US for wider access to Mythos will also be important. Meanwhile, the government’s cautionary stance against premature adoption of foreign AI models like OpenAI and Anthropic in sensitive sectors underscores the emphasis on secure, homegrown technological solutions.