Anthropic's LLM Claude Discovers Cryptanalytic Attacks
Anthropic's language model Claude has discovered new cryptanalytic attacks on post-quantum signature scheme HAWK and 7-round AES-128, but experts say they pose no threat to current security standards. The breakthrough reduces the estimated security level of HAWK-512 from 128-bit to at most 108 bits. The findings highlight the potential of LLMs in cryptography research, but also emphasize the need for formalizing attack techniques.
Key points
- Anthropic announced new cryptanalytic attacks discovered using their LLM Claude on HAWK, a post-quantum signature scheme candidate, and 7-round AES-128.
- The attacks reduce the estimated security level of HAWK-512 from 128-bit to at most 108 bits.
- Experts say the attacks pose no threat to current security standards, but highlight the potential of LLMs in cryptography research.
- Anthropic believes further exploring LLM-assisted cryptography research is interesting, particularly in cases where attacks are computationally intractable to implement.
- The findings emphasize the need for formalizing attack techniques used in the cryptographic community.
Anthropic's LLM Claude has made a significant breakthrough in cryptography research by discovering new cryptanalytic attacks on post-quantum signature scheme HAWK and 7-round AES-128. The attacks, which were announced on July 28th, reduce the estimated security level of HAWK-512 from 128-bit to at most 108 bits.
While the attacks pose no threat to current security standards, experts say they highlight the potential of LLMs in cryptography research. Anthropic believes that further exploring this direction is interesting, particularly in cases where attacks are computationally intractable to implement.
The findings also emphasize the need for formalizing attack techniques used in the cryptographic community. As Anthropic notes, this would allow researchers to better understand and develop more secure cryptographic systems.
The impact of this breakthrough is still being assessed, but it is clear that LLMs have the potential to revolutionize cryptography research. As the field continues to evolve, it will be interesting to see how LLMs are used to develop more secure and efficient cryptographic systems.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.