Something fundamental has shifted in cybersecurity, and while a few nations are beginning to grasp the implications, most, particularly India, remain dangerously unprepared for what comes next.

Last week, Anthropic revealed that its latest Artificial Intelligence (AI) system can autonomously discover, chain, and weaponise software vulnerabilities at machine speed, far beyond anything human teams can achieve. It identified thousands of high-severity zero-days across major operating systems and browsers and even uncovered a flaw in OpenBSD’s TCP stack that had survived 27 years of audits and stress testing.

This rewrites the rules of cyber conflict by shifting attacks from tricking humans into clicking malicious links to machines that can independently find and exploit vulnerabilities. The entire lifecycle of an attack, the reconnaissance, exploitation, and persistence, can now run as a continuous, automated process operating faster than humans can respond.

Anthropic fully grasped the explosive implications of its breakthrough and deliberately chose not to release the system publicly. Instead, it created Project Glasswing, a highly exclusive, tightly controlled initiative that grants access only to a privileged circle of America’s most critical institutions: Apple, Google, Microsoft, AWS, Nvidia, and JPMorgan Chase. The Pentagon and Wall Street were promptly briefed.

Tellingly, even though this same technology poses an equally grave — if not greater — risk to India’s banks, tech giants, and critical infrastructure,........

© hindustantimes