AI is changing how businesses operate. However, it is also changing how cyberattacks happen. The Indian government has now stepped in with a clear warning to banks. As more powerful AI systems are developed, the risk is no longer just about new types of attacks. Instead, it is about how quickly existing threats can evolve. At the center of this concern is a model that has not even been released publicly yet.

Breakdown:

The concern revolves around Anthropic’s advanced AI model, Claude Mythos, which is designed to handle complex cybersecurity tasks. Unlike earlier systems, this model can identify software vulnerabilities, analyse systems, and even generate exploit pathways at scale. While it is currently being tested in a controlled environment, early reports suggest that it can outperform humans in certain cybersecurity functions.

As a result, the Indian government has asked banks to strengthen their cybersecurity preparedness. In a meeting led by Finance Minister Nirmala Sitharaman, lenders were advised to take pre-emptive steps to protect systems, secure customer data, and monitor threats more closely. The concern is not just theoretical. There are already indications that unauthorised access occurred during testing, raising questions about how such systems could be misused if they fall into the wrong hands.

What makes this shift more significant is the speed at which threats can now emerge. Traditionally, vulnerabilities in systems could take weeks or even months to identify and exploit. However, with AI, that timeline is shrinking rapidly. Experts suggest that tasks which once required specialised teams can now be executed in hours, and at a much larger scale. This means attackers can test multiple pathways simultaneously, making fraud more targeted and harder to detect.

At the same time, the challenge is not entirely one-sided. The same AI capabilities that can be used to exploit systems can also be used to defend them. This creates a new dynamic where cybersecurity becomes a race between AI systems. As a result, institutions that adopt AI-driven security faster may gain an advantage, while those that lag behind could face greater risk.

Why this matters:

This marks a shift in how cyber risk is evolving in India’s financial system. The issue is no longer just about preventing attacks but about keeping pace with how quickly they can develop. As AI reduces the time needed to identify and exploit vulnerabilities, banks and institutions may struggle to respond fast enough. At the same time, this increases the importance of proactive security measures rather than reactive ones.

The Big Picture:

More broadly, this reflects a global shift in cybersecurity. Governments and regulators across countries are beginning to recognise that AI is fundamentally reshaping the threat landscape. The challenge is not just technical but structural, as traditional defence systems were not designed for machine-speed attacks. As a result, the future of cybersecurity is likely to be defined by AI versus AI, where automated defence systems must match the speed and scale of automated threats.

The Crunch:

AI is not just making businesses faster. It is making cyber threats faster too. In this environment, the real risk is not a new kind of attack, but how quickly existing ones can scale beyond control.