AI is having a moment, but it's not a danger to cryptography — yet

Anthropic published a July blog post showing its Mythos model found a way to weaken the digital signature scheme HAWK, prompting the algorithm's research team to pull it from NIST's standardisation assessment scheme after fixes made it "uncompetitive." HAWK researcher Léo Ducas told The Stack that "the notion that AI will break all cryptography or that we would need some kind of post-AI cryptography is a take that doesn't make sense to me," rejecting panic over AI's impact on cryptography.

When it comes to offensive cybersecurity tasks, the impressive capabilities of LLMs are well documented, especially since the launch of Anthropic’s Mythos model. But the lab has also claimed Mythos can find mathematical flaws in post-quantum algorithms: Does this signal a potential disaster for cryptography, or is it another attempt to consolidate power disguised as a “warning?” Anthropic published a blog in July https://www.anthropic.com/news/discovering-cryptographic-weaknesses?ref=thestack.technology showing that Mythos had found a way to weaken the digital signature scheme HAWK, leading the research team behind the algorithm to pull it from NIST’s standardisation assessment scheme https://groups.google.com/a/list.nist.gov/g/pqc-forum/c/2r2u6SbHun4/m/0 I2KOZ CQAJ?ref=thestack.technology after finding that fixes made it “uncompetitive.” That decision sparked fresh conversations about the impact of AI on our security systems and whether the industry should add the fear of AI-powered attacks on cryptography to their concerns alongside the impending quantum threat. However, HAWK researcher Léo Ducas sees no need for panic. “The notion that AI will break all cryptography or that we would need some kind of post-AI cryptography is a take that doesn't make sense to me,” he told The Stack in a recent interview.