NewsBite analysis · AI, Security
The Decoder
Anthropic says its Mythos model found vulnerabilities in cryptographic algorithms that secure the internet
Anthropic's Mytho model has discovered significant vulnerabilities in two widely used cryptographic algorithms. The model improved an attack on HAWK, a post-quantum signature scheme, and found a new attack on a reduced version of AES, highlighting the potential of AI in cryptanalysis. Although these vulnerabilities do not affect current systems in use, they indicate how AI might challenge established security frameworks. The implications of these findings could influence future cryptographic standards and practices, especially as quantum computing evolves.
MonitorConfidence: MediumShort term horizonCybersecurity risk: High
Published 28 Jul 2026, 20:10 UTC3 min read
Visit original article ↗Lead image — The Decoder
Summary
TL;DR
- Mythos found an improved attack on the HAWK cryptographic scheme.
- A new attack on a 7-round reduced version of AES was discovered.
- The findings demonstrate AI's potential in challenging security assumptions.
- Neither vulnerability currently impacts systems in operation.
- Mythos' research involved both human researchers and semi-autonomous AI agents.
- The attack on AES improved upon previous known attacks significantly.
Interesting facts
- Mythos developed an innovative fingerprinting method called 'Möbius Bridge'.
- Human researchers not specializing in cryptography contributed to the project.