Claude Mythos AI Model Pokes Hole in Leading Post-Quantum Cryptography Candidate

Crypto majors are green ahead of the FOMC meeting, but it's not just about market sentiment. A recent development in post-quantum cryptography has left experts scrambling to patch vulnerabilities in a leading digital signature system.

Anthropic's powerful AI model, Claude Mythos Preview, discovered two previously unknown attacks on crypto algorithms, including HAWK, a scheme competing to become a U.S. federal standard. HAWK was designed to survive quantum computers and proved a transaction came from the owner without exposing their private key.

Claude found a symmetry in HAWK's math that allowed it to steal its smallest key with roughly 67 million times less work than previously thought, dropping the cost from 2^64 operations to 2^38. This means patching HAWK would involve doubling its key sizes, essentially negating its compact and fast signing advantages.

The incident has raised concerns about post-quantum security in crypto, as experts had assumed that post-quantum schemes replacing today's cryptography are themselves sound. Claude Mythos' discovery pokes a hole in the leading candidate before it even shipped, leaving the industry racing to test its own code with the powerful tool.