Anthropic's artificial intelligence model, Claude, achieved what leading global cryptographers couldn't for two years. In just 60 hours of work, the AI identified hidden mathematical patterns in two encryption systems, casting doubt on their reliability. These are protocols designed to protect messages, logins, and banking data from outsiders.
It's important to emphasize: current security standards used by millions of people remain unaffected. However, the fact that a machine discovered what experts overlooked changes the rules of the game in cryptography.
Blow to the Post-Quantum Protocol HAWK
HAWK is a promising digital signature system developed with the threat of quantum computers in mind. It was long considered secure. However, Claude found a hidden pattern in its mathematical foundation that radically reduces security.
According to Anthropic's analysis, cracking the smallest HAWK key now requires only 2^38 steps instead of the expected 2^64. This means a reduction in computational cost by approximately 67 million times. A solution exists, but it's extremely inconvenient: the key size must be doubled, negating HAWK's main advantage—its efficiency.
Attack on AES: The "Möbius Bridge"
The second object of study was AES (Advanced Encryption Standard), the primary encryption standard protecting most internet traffic since 2001. Claude didn't crack the full 10-round AES but analyzed its simplified 7-round version.
The AI found a non-standard attack method, named the "Möbius Bridge." The essence is that one of the selection stages turned out to be redundant. This accelerated the most effective attack on the simplified version by 200–800 times. Claude took three days for the full analysis, after which two experts spent nearly a month verifying the results.
New Reality: Humans Can't Keep Up with AI
Anthropic directly points to a key problem: humans can no longer keep pace with the model's speed. The main delay now occurs not at the discovery stage but at the result verification stage. This echoes the situation in 2022, when researchers cracked another post-quantum algorithm, SIKE, on a regular laptop in an hour.
It's important to note that Bitcoin and Ethereum are built on different cryptographic principles, and these attacks do not directly affect them. However, time is working against us. Google plans to transition to new standards by 2029, while U.S. government agencies have a deadline of 2031.
What's Next?
Now, NIST must evaluate the public attack on an algorithm that was still being considered for implementation. Anthropic has indicated that not all results have been published. In the near future, the company will hold a special seminar with scientists.
Cryptalist Expert Opinion: This case is not just a technical discovery. It's a demonstration that AI is beginning to surpass humans in tasks requiring intuition and sifting through millions of options. The question is no longer "can AI crack ciphers?" but "will we be able to implement new standards before malicious actors do?" The cryptocurrency and DeFi market needs to prepare for a new security paradigm today.