---
title: "Discovering Cryptographic Weaknesses with Claude Mythos"
source: "Simon Willison 博客（Atom）"
source_tier: "t2"
url: "https://simonwillison.net/2026/Jul/28/discovering-cryptographic-weaknesses-with-claude/#atom-everything"
permalink: "https://feed.astralor.com/items/itm_019b9ac466bc0a2c83b44a4c"
published: "2026-07-28T22:45:37.000Z"
score: 74
channel: "ai"
tags: ["AI Security Research","Anthropic","Claude Mythos","Cryptography","Prompt Engineering"]
---

# Discovering Cryptographic Weaknesses with Claude Mythos

## AI commentary

> This is an exciting experiment—it proves that large models can persist for hours on truly hard research problems and produce meaningful mathematical discoveries. For AI safety experts, cryptographers, and prompt engineers, the core value lies in the design philosophy behind those seemingly casual prompts: not giving the model answers, but forcing it to think.

## At a glance

Anthropic researchers used the Claude Mythos model, guided by carefully crafted prompts (including spelling mistakes), to discover mathematical flaws in HAWK and a weakened version of AES. While these findings have no practical impact on current systems, the experiment demonstrates the potential of LLMs in advanced research. The model ran for 60 hours, costing approximately $100,000 in API usage, with the main human intervention being encouragement to "not give up" and "find something worth publishing." The paper and code are publicly available.

## Key points

- Claude Mythos successfully identified mathematical weaknesses in HAWK and a weakened version of AES.
- Researchers guided the model with repeated prompts like "don't give up" and "find something worth publishing."
- The experiment ran for 60 hours with an estimated API cost of $100k, and the results pose no practical threat to current systems.

## Full text

Discovering cryptographic weaknesses with Claude The best part of this article (here's the repo) about how Anthropic researchers used Claude Mythos to find mathematical flaws in both HAWK and a weaker version of AES ("neither of these results has a practical impact on today’s computer systems") is the prompts that they shared, spelling mistakes included: the models tend to think it is impossible to solve so they don't try they need a good amount of prompting. why not do aes-128 r7? the whole point is to find something better than existing approaches. no again the goal is that we have highly inteligent model as good top researcher, we want to find new attacks no we don't want to change the targets [...] agian we need to find something that worth publishing again we are not looking for low hanging fruit, we want proper research to find genuinly hard findings. Mythos Preview worked for 60 hours in total (~$100,000 in estimated API cost) and the main human interventions were to encourage it not to give up and "find something that worth publishing". Via Hacker News Tags: ai, prompt-engineering, generative-ai, llms, anthropic, claude, ai-security-research, claude-mythos-fable

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_019b9ac466bc0a2c83b44a4c