AI news story
Anthropic says its Mythos model found vulnerabilities in cryptographic algorithms that secure the internet
Anthropic's Claude Mythos Preview found weaknesses in key cryptographic algorithms, including a better attack on HAWK, a post-quantum signature scheme that human experts had reviewed for more than two years. The model found it in just 60 hours at an
Editor's take
Anthropic's AI model, Mythos, has identified previously unknown vulnerabilities in widely used cryptographic algorithms, including a novel attack on the HAWK post-quantum signature scheme.
This discovery is significant because it highlights the potential for advanced AI to find flaws in systems designed to protect sensitive data, impacting internet security and post-quantum cryptography research. The fact that Mythos found this weakness in just 60 hours, after human experts had scrutinized HAWK for over two years, suggests a new era in security auditing where AI could become an indispensable tool for identifying emergent threats.
Future developments to monitor include whether Anthropic plans to release Mythos or similar AI security tools, and if other AI labs, such as Google DeepMind or OpenAI, will develop comparable capabilities. The ability of AI to rapidly discover algorithmic weaknesses could necessitate a faster pace of cryptographic algorithm development and deployment.
Signal score: 4
This event was corroborated by 50 independent sources. The signal score weighs cross-source corroboration, recency, source weight and topic salience. How we rank stories.
Original reporting
This story summarises reporting published by The Decoder. Read the original article at The Decoder.