Claude Mythos – The thin line between cybersecurity and hacking

Claude Mythos from Anthropic has emerged as a serious risk for global banks. The fact that the Finance Minister summoned bank chiefs and RBI officials for a discussion, underlines the gravity of the subject.

What is Claude Mythos all about?

If you have been tracking the sharp fall in Indian IT stocks, the reason was a new platform developed by Anthropic of the US. The same Anthropic has now launched Claude Mythos. On the face of it, Claude Mythos appears to be an innocuous piece of AI-based software that is being currently tested by a closed-user group. The scary part is the claim made by Anthropic about Claude Mythos. Apparently, Claude Mythos can outperform humans at hacking and cybersecurity risks. To put it bluntly, this could be a ticking time bomb. It is meant to make cybersecurity frameworks tighter and hack-proof. However, in the process, this is a bit like giving the keys of the vault, voluntarily to the burglar.

Early tests confirm the claims

Researchers who had done early tests on Claude Mythos have confirmed that the AI-powered software was perfectly capable of efficiently hacking into security systems. In short, it could create potent computer risks, especially for sectors like banking and financial services which deal with mountains of confidential customer data and payment systems. Apparently, Claude Mythos has the power to locate bugs that were lying dormant in legacy code. Eventually, whether such knowledge is used to address the risk or to exploit the risk is a matter of debate. In fact, Claude Mythos is so powerful that the company was willing to work with the US government to tighten defences.

How will outcomes be controlled?

As of now, nobody has an answer to that question. Early tests threw up thousands of high-severity vulnerabilities across major operating systems and web browsers. For now, Claude Mythos has been administered in a limited environment. However, the fear is that it may not remain limited for too long. It is a bit like a nuclear bomb; a great idea if in responsible hands, but disastrous if in riskier pockets. In fact, Claude Mythos can get into decades old code and highlight bugs. Now comes the challenge on how responsibly these bugs are handled. It is this “Unknown Unknown” that is giving sleepless nights to Indian financial regulators. Nobody knows how virulent the downsides could be!

Are tech researchers crying wolf?

At least, the Indian government does not think so. They see distinct risks to Indian financial system cybersecurity. That means; millions of customers, their confidential data, and their funds could be substantially vulnerable. The Finance Minister has been repeatedly warning the financial system about this risk. After all, an unknown risk is always more potent than known risks. In the technology world; there is a thin line between a security expert and a security hacker. When the power of AI is in your hand, the temptation to become a rogue researcher is quite high. That is exactly what the Indian finance minister is worried about; and she is perfectly justified!