An event that could upend the perception of data security has occurred in the world of cryptography. Anthropic's artificial intelligence, Claude, identified critical vulnerabilities in two encryption systems during a study. These are mathematical algorithms that currently protect our messages, logins, and bank transactions from prying eyes.

It is important to emphasize: the current encryption standards used by millions of people are not yet in danger. But the significance of this discovery lies elsewhere—the machine found what leading global experts, who had worked on these algorithms for years, had overlooked.

HAWK: 67 million times faster

The first target is the HAWK protocol, a digital signature originally designed with the threat of quantum computers in mind. For two years, experts found no flaws in it. Claude needed only 60 hours to notice a hidden pattern in HAWK's mathematical foundation.

The result is impressive: according to Anthropic's calculations, cracking the smallest HAWK key is now possible in 2^38 steps instead of the assumed 2^64. Computing resource costs have been reduced by approximately 67 million times. There is a solution to the problem, but it is radical—doubling the key size, which effectively nullifies the algorithm's main advantage.

AES: "Möbius Bridge" accelerates attack by hundreds of times

The second target is AES (Advanced Encryption Standard), the primary standard encrypting most internet traffic since 2001. Claude did not crack full AES. The study was conducted on a simplified version with 7 rounds instead of the standard 10.

However, the method found by the AI, called the "Möbius Bridge," proved truly innovative. The essence is that one of the key selection stages turned out to be redundant. As a result, the most effective attack was accelerated by 200–800 times. It took Claude three days to work alone. Then, two specialists spent nearly a month double-checking the obtained result.

A new reality: humans can't keep up with AI

Now, the main delay occurs at the verification stage of the result. Anthropic's researchers note a paradox: humans can no longer keep pace with the model's speed. This is reminiscent of the 2022 story when researchers cracked another post-quantum algorithm, SIKE, in about an hour on a regular laptop, but here the speed of hypothesis generation is fundamentally different.

It is important to clarify: Bitcoin (BTC) and Ethereum (ETH) operate on different principles. These attacks do not affect them. But time is of the essence. Google plans to transition to new standards by 2029, and U.S. government agencies have a quantum deadline of 2031.

The National Institute of Standards and Technology (NIST) now must evaluate a public attack on an algorithm that was still being considered for implementation. Anthropic indicated that not all attack results have been published yet. The company plans to hold a special seminar with scientists in the near future.

Expert comment: This is a wake-up call. If before we debated whether a quantum computer could crack cryptography, now the question is different: won't an ordinary AI do it much sooner? The security standards we considered unshakable could collapse not under the onslaught of quantum physics, but due to the mathematical intuition of a neural network.