The level of concern is visible at the highest levels—US Federal Reserve Chair Jerome Powell met American bankers earlier in April, while India’s Finance Minister Nirmala Sitharaman and IT Minister Ashwini Vaishnaw met Indian bankers last week to discuss the issue.
What exactly is Mythos and how is it different from existing AI models?
It is not a general-purpose chatbot. Think of it as a super-expert version of AI tools like ChatGPT or Gemini —not just a smarter chatbot.
Mythos is designed for cybersecurity and coding, with the ability to scan systems, detect weaknesses and fix them. But those same capabilities can also be used to break into systems.
What makes it different from earlier AI models is its level of independence. While older systems require step-by-step human input, Mythos can analyze a large system, identify vulnerabilities and execute multi-step actions on its own. It can scan entire operating systems or browsers in hours—tasks that would typically take teams of experts weeks or months.
Its most significant capability is identifying “zero-day” vulnerabilities—bugs that developers are not aware of. These are among the most critical risks in cybersecurity.
Mythos has already identified multiple such vulnerabilities across widely used systems. It discovered 271 issues in Mozilla Firefox in a single cycle. In OpenBSD, it uncovered a 27-year-old flaw that could allow a remote attacker to trigger a denial-of-service crash. In FFmpeg, used by platforms like YouTube and Netflix, it identified a 16-year-old vulnerability that had passed millions of tests. It has also flagged vulnerabilities in the Linux kernel, which is widely used across devices and cloud networks.
What’s worrying banks and regulators?
Such an advanced model creates both opportunity and risk. On one hand, Mythos allows companies to identify and fix vulnerabilities before they are exploited. On the other, if such a system becomes widely accessible, it could allow individuals with limited expertise to carry out sophisticated cyberattacks.
That is why Anthropic has not released Mythos publicly. Access has been restricted to a group of vetted organizations, including major technology firms such as Microsoft, Google, Nvidia, Apple, CrowdStrike and Cisco, as well as institutions like the Linux Foundation and global banks including JPMorgan, Citigroup, Goldman Sachs and Bank of America.
These institutions are testing the system under high-security conditions to identify and fix vulnerabilities in critical software before similar capabilities become widely available.
What is India doing about the potential risks?
In India, the risks are being closely assessed, particularly in the financial sector and digital public infrastructure such as Aadhaar and Unified Payments Interface (UPI). The Ministry of Electronics and Information Technology is engaging with global stakeholders to explore access for domestic systems.
Following high-level discussions in April 2026, a panel involving the State Bank of India Chairman and the Indian Banks’ Association has been set up to assess risks and prepare a coordinated response for the banking system.
The issue is not limited to India or banking. Critical sectors such as aviation, healthcare, defense and energy could also be affected.
As more advanced AI systems emerge, the challenge is not just about what they can do, but how access to them is controlled and used.
