Claude Mythos AI raises cyber risk: Govt asks MSMEs to stay alert

India’s cybersecurity agency has issued a fresh warning, and this time, the concern isn’t just hackers, it’s AI-powered attacks. The Indian Computer Emergency Response Team (CERT-In) has flagged a new wave of cyber risks linked to advanced AI systems like Claude Mythos. These tools, which were originally built for research and development, are now raising concerns for how easily they can be misused. Also Read: Why Google invests billions in rival Anthropic amid growing AI race?

For businesses, especially MSMEs, this is not something to ignore. Also Read: ChatGPT 5.5 vs Claude Opus 4.7: Which model should you trust?

Why is the government alerting about risks of AI models?

At the centre of this advisory is a new generation of AI models that can do more than just generate text or code. These systems can analyse large software systems, find vulnerabilities, and even simulate attacks, all with very little human input. In simple terms, tasks that once required skilled cybersecurity teams can now be done faster, cheaper, and at scale using AI. Also Read: AI race heats up as Google eyes $40 billion bet on rival Anthropic

That’s where the real concern begins. CERT-In notes that such tools can scan networks, identify weak points, and even plan multi-step attacks automatically. This significantly lowers the barrier for cybercriminals.

Why MSMEs should be more careful

Large enterprises usually have dedicated security teams. But MSMEs often don’t have that luxury. And that makes them easier targets. According to the advisory, AI-driven attacks could lead to data breaches, financial fraud, identity theft, service disruptions, and unauthorised system access. There’s also the risk of “cascading impact”, where one breach affects multiple connected systems.

For smaller businesses that rely heavily on digital tools, even a single incident can cause serious damage.

Risk of smarter scams

One of the more worrying parts is how AI is changing phishing and scams. These are no longer basic spam emails with obvious mistakes. AI can now create convincing emails, realistic voice calls, and even deepfake videos. All of this makes it harder to tell what’s real and what’s not.

So the usual “be careful online” advice now needs to be taken more seriously.

What the government is asking companies to do

The advisory doesn’t just warn, it also suggests what needs to be done. For organisations:

• Monitor systems more actively
• Apply security updates quickly
• Use multi-factor authentication (MFA)
• Track unusual activity patterns
• Avoid outdated software and systems

For MSMEs:

Add Techlusive as a Preferred Source

• Keep apps and systems updated
• Use strong passwords and MFA
• Avoid using unverified AI tools
• Train employees about cyber risks

And for individuals, the advice is simple, verify before you trust.