India, tech firms examine cyber risks linked to Anthropic's next-generation Mythos AI

India is assessing the cybersecurity risks posed by Anthropic's advanced AI model Mythos, with government agencies and major technology firms examining vulnerabilities in crucial software systems used across banking and public infrastructure.

According to a Bloomberg report, India has started testing some of its most sensitive public-facing financial and government software systems to understand how vulnerable they could be to Mythos, Anthropic's next-generation artificial intelligence model.

The report stated that the Indian IT companies Infosys and Tata Consultancy Services (TCS) are among firms carrying out tests in secure environments. Infosys is reportedly working on identifying and patching vulnerabilities in its widely used Finacle banking software.

India's cybersecurity agency CERT-In is testing key digital infrastructure, including Aadhaar-related systems and government login platforms, Bloomberg reported, citing officials who are familiar with the matter.

What is Mythos AI and why is it causing concern?

Anthropic recently revealed Mythos as its most powerful AI model to date. According to a report by The Indian Express, the company has deliberately restricted public access to the model because of concerns over its extraordinary capacity to autonomously identify serious software vulnerabilities.

Anthropic has said Mythos can detect flaws across major operating systems, browsers and widely used software infrastructure. The company fears such capabilities could potentially be misused for cyber attacks if released widely without any safeguards.

The Indian Express reported that Anthropic is in talks with several governments, including India, on securing critical infrastructure such as telecom, banking, and energy systems against emerging AI-linked cybersecurity risks.

"There is a belief within Anthropic that allied democracies need access to defence capabilities against powerful AI models," a source stated to The Indian Express.

The discussions between Anthropic and Indian authorities were reportedly started through India's Ministry of External Affairs.

Indian government steps up vigilance

Concerns related to Mythos have already reached the highest levels of the Indian government.

According to the Indian Express, Union Finance Minister Nirmala Sitharaman recently headed a high-level meeting to review potential risks to India's financial sector from advanced AI systems capable of weaponising software vulnerabilities.

In a statement, the Finance Ministry stated that the emerging threat from AI models was "unprecedented" and required a "high degree of vigilance, preparedness, and better coordination" among financial institutions and banks, as quoted by The Indian Express.

While the ministry did not clearly name Mythos, the discussion were related to concerns related to the model's capabilities.

Global debate over Mythos threat

Despite ongoing global concern, some cybersecurity experts believe fears around Mythos may be overstated.

According to a report by Reuters, security researchers acknowledge that Mythos represents a major advancement in vulnerability detection, but many argue that AI-assisted cybersecurity testing has existed for years.

"I think there's a really big communication gap between practitioners and policy makers," says Issac Evans, founder of software security firm Semgrep, stated to Reuters.

Reuters reported that Mythos is capable of identifying software flaws faster and with simpler prompts than earlier AI systems. It is potentially lowering the barrier for cybercriminals. However, experts also mentioned that finding vulnerabilities is the only first step whereas validating and exploiting them remains far more complex.

Cisco executive Anthony Grieco mentioned to Reuters that Mythos can help defenders scan large amounts of code more quickly and reduce false positives. It allows cybersecurity teams to focus on major threats more efficiently.