Mythos attack on 3rd-round PQC algorithm candidate puts it out of commission
An AI model developed by Anthropic, called Mythos, helped discover a cryptographic weakness in HAWK, a digital signature scheme designed to resist attacks from quantum computers, prompting its developer to withdraw it from consideration as a US government standard. HAWK had already passed two rounds of evaluation by the National Institute of Standards and Technology (NIST) as part of its post-quantum cryptography (PQC) standardisation process, and was undergoing a third round intended to catch exactly this kind of flaw. The finding is notable because it suggests AI systems may now be capable of contributing meaningfully to cryptanalysis, a field with major implications for future digital security.
Working with roughly 60 hours of effort and about $100,000 in computing costs, an Anthropic researcher without cryptography expertise used Mythos to improve upon the best-known existing attack against HAWK, effectively halving its key strength. HAWK's security rests on the presumed difficulty of the Lattice Isomorphism Problem, believed to resist quantum attacks, and Mythos produced a previously unknown method for exploiting so-called automorphism symmetries within it. The weakness could theoretically be mitigated by doubling HAWK's key size, but the resulting performance cost has made the algorithm less attractive and contributed to its withdrawal. Anthropic noted similar, more limited progress was made against the widely used AES cipher, though experts caution the results are incremental, tested against deliberately weakened "challenge" versions of the systems, and do not threaten any cryptography currently in real-world use.
In short
- Anthropic's Mythos AI helped break a candidate quantum-resistant algorithm, HAWK
- HAWK's developer withdrew it after Mythos halved its key strength
- Mythos also made smaller progress attacking the AES cipher
- Experts stress current encryption in use remains unaffected for now
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The discovery supports using powerful AI tools in cryptographic review: they may help researchers find subtle weaknesses that conventional evaluation misses, including in systems intended to withstand quantum attacks. HAWK was still in a review round designed to uncover such flaws, and withdrawing it before adoption can help protect future standards and give developers useful evidence about the limits of their designs.
The case against
The result may be overstated if treated as evidence that AI can break widely used cryptography: the reported attack was incremental, tested against a challenge version, and did not threaten deployed systems. A costly attack on a candidate that had already passed earlier reviews also raises questions about how much confidence to place in AI-assisted findings without independent reproduction, and whether the performance trade-off from enlarging keys makes HAWK’s withdrawal the right response.
Explained simply
Computers using quantum mechanics could theoretically break the encryption that currently protects digital communications. Governments and technology companies are developing new encryption systems designed to resist quantum computers, a field called post-quantum cryptography. This is being done as a precaution: quantum computers powerful enough to pose this threat do not yet exist, but such systems would be needed if they ever do, particularly because data encrypted today might be decrypted in the future.
The US government's National Institute of Standards and Technology (NIST) is leading an international effort to evaluate and standardise the most promising new encryption methods. An algorithm called HAWK, which encodes security within mathematical structures, had successfully passed through two rounds of this rigorous assessment process. It was undergoing the final round, where candidates face intense scrutiny specifically designed to uncover weaknesses that earlier testing might have missed.
An artificial intelligence system developed by Anthropic, called Mythos, discovered a previously unknown weakness in HAWK that reduces its encryption strength. The researcher using Mythos lacked expertise in cryptography, suggesting that artificial intelligence might now contribute to discovering flaws in encryption systems, work that has traditionally required specialist expertise. Though experts note the attack worked against a deliberately weakened test version and poses no immediate threat to real-world systems, the finding suggests artificial intelligence may be capable of making meaningful contributions to this field.