Anthropic Claude Mythos Preview Finds Cryptographic Weaknesses
Anthropic researchers applied the model to identify flaws in secure digital signature schemes and symmetric ciphers.
TLDR
Anthropic published research showing its Claude Mythos Preview model assisted researchers in uncovering vulnerabilities in cryptographic algorithms that protect private data. The work focused on a highly secure digital signature scheme for identity verification and a well-known symmetric cipher for encryption. Official company statements and linked sources confirm the findings, highlighting AI's role in probing mathematical methods for data privacy. Reactions from engineers and observers noted potential broad impacts on cybersecurity infrastructure and the need for stronger protections.
Combined views
3.2M
24 Sources, first seen ago
Sources
BD FT 
Florian Tramèr@florian_tramerBy running models *slightly* longer than in our recent benchmark evaluation (one billion tokens and >$100k ) Mythos found some impressive attacks on cryptographic schemes https://twitter.com/AnthropicAI/status/2082153297670992134
- likes: 84
- replies: 3
- bookmarks: 14
- reposts: 6
ND Nick Dobos@NickADobosHow long before we get another hugging face incident but this time its Claude stealing bitcoin? https://twitter.com/anthropicai/status/2082153297670992134
- likes: 29
- replies: 7
- bookmarks: 1
- reposts: 0
SI signüll@signullloh my god. the fundamental principles upon which almost all of cyber security, hell even some physical security itself is built are rapidly breaking down. if this is true, & a model was able to find a way to break symmetric cryptography every piece of infrastructure is…
- likes: 2K
- replies: 160
- bookmarks: 982
- reposts: 77
XL xlr8harder@xlr8harderYou know, I'm something of a prompt engineer, myself. https://twitter.com/Sauers_/status/2082171683645817193
- likes: 60
- replies: 4
- bookmarks: 2
- reposts: 0
XL xlr8harder@xlr8harderI'm annoyed that in their post they say they are releasing "a document containing Claude’s chain of thought" but what they've actually shared has been completely rewritten and narrativized. https://twitter.com/AnthropicAI/status/2082153297670992134
- likes: 149
- replies: 5
- bookmarks: 16
- reposts: 2