The lock that will always be broken
For decades, we have treated encryption as a mathematical fortress, but we are entering a new era – Post-AI Cryptography.
A post-quantum encryption scheme, HAWK, has been exposed to a significant vulnerability by an AI system, Claude Mythos Preview, developed by Anthropic. The AI reduced the effort needed to crack the smallest configuration of HAWK by approximately 67 million times. As a result, Anthropic withdrew their candidate from the National Institute of Standards and Technology's (NIST’s) post-quantum digital-signature standardisation process.
Other post-quantum candidates, such as SQIsign, also raised concerns. Despite this, the three post-quantum standards NIST finalized in 2024 remain unaffected. This incident emphasizes that no algorithm can be considered permanently secure. The tech industry is now using AI to find and repair vulnerabilities before criminals can exploit them.
However, the focus has shifted from merely developing stronger algorithms to using machines to fight machines. The fightback is no longer just about developing stronger algorithms, but about using machines to find security holes before attackers can exploit them. With concerns growing over AI models misbehaving, there is a need for a more holistic approach to security that enables systems to adapt, contain damage, and recover when protections fail.
Written by urgent.news from ITWeb's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
