Here's What Anthropic Found when It Turned Mythos Loose on Encryption Algorithms
Anthropic said the model found flaws in two ciphers after $100,000 runs, including a post-quantum scheme under NIST review.
- On Tuesday, Anthropic announced its AI model, Claude Mythos Preview, discovered novel mathematical flaws in HAWK and a weakened version of the Advanced Encryption Standard , engineering attacks 200 to 1,000 times faster than previous human research.
- Mythos engineered these breakthroughs in 60 hours of semi-autonomous work, each discovery costing approximately $100,000 in API compute, while independently devising a mathematical shortcut it named the "Möbius Bridge."
- Neither result affects software currently in use; the AES attack targets a simplified seven-round test version, while the HAWK vulnerability reduces effective key strength, necessitating larger key sizes to maintain security.
- Ellen Boehm, senior vice president at Keyfactor, said the research proves the NIST PQC evaluation process is effective, though it elevates the need for organizations to maintain visibility over internal cryptographic infrastructure.
- Former National Security Agency general counsel Glenn S. Gerstell questioned whether strong encryption will remain secure as model power grows, as these disclosures follow last week's containment breach where OpenAI agents launched an unauthorized cyberattack.
15 Articles
15 Articles
The Hindu Business Line
LatestlyAnthropic says Claude uncovers cryptographic weaknesses, advances AI-assisted security research
The company noted that while modern cryptographic systems remain highly secure, AI models are increasingly capable of assisting researchers in analyzing complex cryptographic constructions
Anthropic's Claude Mythos Finds Encryption Flaw Following OpenAI Agents' Containment Breach
Anthropic's Claude Mythos Preview has exposed flaws in core encryption standards just one week after OpenAI models escaped sandbox containment and attacked Hugging Face.
Here's what Anthropic found when it turned Mythos loose on encryption algorithms
Anthropic researchers used Claude Mythos Preview to find new weaknesses in two cryptographic methods, the company said Tuesday, including one that is being considered by the National Institute of Standards and Technology for both traditional and quantum computing. In a blog post detailing the work, the frontier AI company called it a “substantial” research advancement, but also emphasized that neither flaw affects software now in use.“The atta…
Claude found mathematical flaws in two cryptographic algorithms that years of expert review missed
Anthropic announced on Monday that Claude Mythos Preview has discovered mathematical weaknesses in two cryptographic algorithms. The first substantially weakens HAWK, a post-quantum digital signature scheme under review by NIST, by cutting its effective key strength in half. The second improves the best-known attack on seven-round AES, the most widely used symmetric cipher, by 200-800x. […] This story continues at The Next Web
Can AI Challenge Quantum Threats? Anthropic Research Raises Fresh Questions for Bitcoin
TL;DR: The Claude Mythos Preview model from Anthropic identified an attack that accelerates the breaking of reduced seven-round versions of the AES algorithm by 200 to 1,000 times. The artificial intelligence tool halved the key strength in HAWK, a candidate signature scheme for NIST’s post-quantum standardization. The Bitcoin network currently uses the ECDSA elliptic curve digital signature algorithm to validate and authorize transactions on…
The AI model "Claude Mythos Preview," which has an extremely high ability to discover software vulnerabilities, has found flaws in two encryption algorithms. Read more...
Coverage Details
Bias Distribution
- 40% of the sources are Center, 40% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
