What was the economic incentive to solve any of the OpenAI problems? Meanwhile ECDLP has had over 3k researchers with 800+ published papers, and it never stopped. There are yearly conferences held to discuss failed attempts at breaking ECDLP, so people can learn from each other and try to break it better. The NSA is the world's greatest hirer of mathematicians. They got caught trying to backdoor Discrete Logarithm systems three times. The last being caught in 2015. After each one, waves of the most competent cryptographers scoured for new attacks. Most things that improve attacks by like 10% are really the same things that make implementations of EC operations faster. Theres armies of people, on the latest frontier models, making their cryptography blazingly fast, like us @ZakuraZcash. You see open research projects constantly improve the state of the art. No human or AI has had any elliptic curve speedup approaching even a 2x, yet folks are freaking out that cryptography is broken? The OpenAI problems being broken is not remotely the same as ECDLP having a real improving attack. Even if there was an attack, its insane to imagine it becomes "Break ECDLP on your laptop". It would go from "impossible" to "possible with all the world's GPU's dedicated for a decade", to "possible with 1 year". Any of those is a real warning shot to go trigger the ECC freeze.