[Cryptography] "Anthropic A.I. Model Finds Flaws in Tough-to-Crack Encryption Algorithms"

[Steven M. Bellovin]

On 31 Jul 2026, at 0:11, Jerry Leichter wrote:

>
>
> It's a faster attack on a weakened version of AES.  Also mentioned is some kind of an attack on a proposed post-quantum protocol (HAWK).
>
> Gift link to the NY Times article about this below.  Obviously the Times is not the place to learn any technical details.  Anyone come across a deeper description?
>
> https://urldefense.com/v3/__https://www.nytimes.com/2026/07/28/us/politics/anthropic-ai-encryption-security-aes.html?unlocked_article_code=1.11A.VL6S.yU2IteGyRauc&smid=url-share__;!!BDUfV1Et5lrpZQ!SNyG80FP5lou0SDtdyL_IR7cqmi99HUbXiHS0sxLVPPf0zetjrArMy4ZB39XfEsXdzzmlxkpn0kPqcfVyw$ -- Jerry
>

See:
https://www.anthropic.com/research/discovering-cryptographic-weaknesses (which links to a draft research paper and another document on Claude's train of "thought".

https://blog.cryptographyengineering.com/2026/07/29/some-notes-about-anthropics-new-results/ (Matt Green's blog post on it)


        --Steve Bellovin, https://www.cs.columbia.edu/~smb