If AI really finds a way to crack existing cryptographic methods, can we still use the assets in our wallets, banks, and brokerage accounts as confidently as we do today? OpenAI recently released a batch of mathematical research results generated by AI. David Friedberg raised a disturbing hypothesis: with so much progress in other areas of mathematics, why haven’t we seen major breakthroughs in cryptography? Could it be that some results are too impactful to be disclosed for now? This is still speculation. The lack of disclosure doesn’t mean it’s already been cracked, and the video doesn’t provide evidence of any breakthroughs. But it’s made me seriously think about what comes next. The security of many public-key cryptosystems is based on the premise that certain mathematical problems are extremely hard to solve. We trust that, with current methods and computational power, attackers can’t find the answers within a reasonable timeframe. (http://pages.nist.gov) But what if AI discovers an effective solution that humans haven’t found? In that case, some systems might face foundational security challenges even before quantum computers become fully mature. For affected digital signatures, the worst-case scenario is this: you didn’t leak your private key, you didn’t click on a phishing link, yet someone else could still produce system-recognized authorizations. And what about banking systems? Bank communications, identity verification, and payment infrastructures also rely on cryptography. The Bank for International Settlements and several central banks are already testing how payment systems can transition to post-quantum cryptography. Migrating such infrastructure inherently requires cross-institutional coordination. Banks still have internal controls, backups, and legal recourse, so it’s not a direct leap to assume all deposits would vanish overnight. What I’m more concerned about is the chain reaction: if a widely used technology suddenly fails, financial institutions might need to suspend some services to verify the authenticity of transactions. Businesses wouldn’t receive payments, individuals couldn’t transfer funds as usual, and the market might start doubting the safety of their money. If panic escalates into mass withdrawals, a cybersecurity crisis could evolve into a liquidity and trust crisis. This is a risk scenario, not something that’s already happening. But it’s something worth preparing for seriously.