InjazAI

Anthropic researchers used Claude Mythos to discover previously unknown mathematical weaknesses in HAWK and a weakened version of AES encryption algorithms, demonstrating AI's ability to conduct advanced cryptographic research at a cost of $100,000.

1 min read

Claude Mythos Discovers Cryptographic Weaknesses: AI Model Outperforms Human Researchers

Discovery Details

FAQ

What is Claude Mythos?

Claude Mythos is an advanced AI model from Anthropic designed for scientific research and complex analysis, used here to discover cryptographic weaknesses.

Do these weaknesses pose a threat to current systems?

No, the results have no practical impact on current systems, but they demonstrate AI's ability to conduct advanced research.

How much did this research cost?

The cost of using Claude Mythos was approximately $100,000 USD for 60 hours of continuous work.

Can this technique be used in the Middle East?

Yes, research teams in the Middle East can use similar models to accelerate cryptography and cybersecurity research, especially with growing AI investments in the region.

Source: Simon Willison (LLM & tools)

AI-assisted content, human-reviewed.