LLMs won't break symmetric crypto
bfswa.blog
August 6, 2026
6 min read
🔥🔥🔥🔥🔥
47/100
Summary
Anthropic's LLM Claude Mythos has discovered key-recovery attacks on HAWK, a post-quantum signature scheme candidate. While the attack is impractical for most versions of HAWK, it significantly reduces the security of the toy version HAWK-256, which has 64-bit security that is nearly breakable.
Key Takeaways
- Anthropic's LLM Claude Mythos discovered a key-recovery attack on the HAWK post-quantum signature scheme, reducing its security level from 128 bits to at most 108 bits.
- An improved key-recovery attack on 7-round AES-128 was found, but it poses no threat to the full 10-round AES-128 implementation.
- LLMs are expected to identify errors in cryptanalytic attacks' complexity estimates and in security proofs, as demonstrated by finding issues in EUROCRYPT 2026 papers.
- The article asserts that LLMs will not break established symmetric cryptographic schemes like AES, ChaCha, and SHA-3 due to their complex and robust design.
Coverage & discussions
Hacker News · bfswa.blog
Community discussions
Hacker News discussion ↗
Historical count: 5 · 5 saved in sample · Historical sample
Analysis ready
Community Sentiment
Positives
- LLMs are poised to supercharge math research, which could lead to breakthroughs that challenge traditional encryption methods — the potential for innovation is enormous.
- LLMs might excel at tedious testing and reasoning tasks in cryptography, where human effort has struggled, suggesting a new era of discovery in security research.
- Recent examples show LLMs identifying basic bugs in encryption libraries that humans overlooked, proving their utility in enhancing crypto implementations.
Concerns
- Skeptics argue that saying LLMs will never break symmetric crypto is overly confident; the effort required is substantial, but past successes in math challenges suggest it's possible.
- Critics point out that the foundational assumptions about the impossibility of reversing cryptographic calculations are shaky, with no guarantees against future breakthroughs.
- Many believe that the extensive human effort in breaking encryption algorithms has yet to yield a viable attack, casting doubt on the notion that LLMs will easily succeed.
