Recent advancements in AI, particularly with language models like Claude Sonnet 4.5, have shown promising results in identifying software vulnerabilities. Anthropic reports that Sonnet 4.5 improves vulnerability detection rates from 2% to 5%, successfully uncovering new flaws in over a third of tested projects. This enhancement was notably showcased during a DARPA AI Cyber Challenge, where teams utilized large language models to create “cyber reasoning systems” capable of analyzing millions of code lines for potential security issues. The development signifies a pivotal shift in cybersecurity, highlighting AI’s growing role in safeguarding digital infrastructure, which is essential for national security and defense strategies, particularly for countries like India that are increasingly investing in advanced technology to bolster their cybersecurity resilience.