Anthropic's AI security model Mythos uncovered critical weaknesses in HAWK, a third-round post-quantum cryptography (PQC) candidate, leading to its withdrawal from NIST standardization. The attack reduced the computational effort needed to break both HAWK and weakened AES variants, highlighting potential advances in cryptanalysis despite not compromising current real-world systems.
Background
NIST has been evaluating PQC algorithms to prepare for quantum computing threats, with HAWK being one of the finalists in the third round. Anthropic's use of an AI model to find flaws represents a novel approach to cryptographic vulnerability assessment.
- Source
- Ars Technica
- Published
- Jul 30, 2026 at 06:07 AM
- Score
- 7.0 / 10