DEVOURED
Anthropic AI Model Finds Flaws in Tough-to-Crack Encryption Algorithms
Anthropic's Claude Mythos Preview autonomously identified flaws in an AES-based encryption algorithm 200 to 1,000 times faster than prior human-led research.
What: Anthropic researchers used Claude Mythos Preview to find vulnerabilities in a simplified AES encryption model. The model independently devised the attack method over one week, which two human researchers took nearly a month to verify.
Why it matters: This demonstrates that LLMs can now perform complex, multi-step cryptographic analysis that historically required months of human expertise, potentially accelerating both defensive and offensive security research.
Decoder
- AES (Advanced Encryption Standard): A symmetric block cipher used to encrypt sensitive data, widely adopted for protecting web traffic and file systems.
- Claude Mythos Preview: A specialized model version from Anthropic optimized for complex reasoning and technical research.
Original article
Claude Mythos Preview found flaws in a watered-down version of an algorithm for Advanced Encryption Standard (AES), a ubiquitous protocol that protects web traffic, wireless networks, data storage, and many other technologies. Anthropic claims the assault was 200 to 1,000 times faster than what previous human research had managed to do. The attack was devised almost entirely autonomously - the model needed aid as it originally believed the problem was impossible. Claude Mythos Preview worked on the problem for about a week before engineering its attack, and two human researchers worked for nearly a month to verify that the method appeared correct.
Explore all 56 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi