Common Knowledge Base Association (CKBA) (@ckba_build): You may have noticed that Crypto Twitter has spent much of this week arguing about whether AI could break the cryptography most crypto wallets rely on…

You may have noticed that Crypto Twitter has spent much of this week arguing about whether AI could break the cryptography most crypto wallets rely on. You may also have struggled to follow the elliptic curves, ECDSA, lattice-based commitments and P = NP stuff. That makes you completely normal. Fear not: nothing has been broken, as far as anyone knows. Here is what happened, what is still unknown, and why it matters. On Tuesday, OpenAI released 722 mathematical manuscripts, written by an internal model it has not made public, covering hundreds of open problems. Mathematicians are still working out what they represent. Fewer than half have computer-checked main results, and a computer check confirms the logic, not the significance. Three papers were withdrawn within a day over a sign error. OpenAI has not claimed an attack on wallet cryptography. One result nevertheless drew serious attention: claimed progress toward the Riemann hypothesis, a puzzle about prime numbers unsolved since 1859. Rutgers mathematician @AlexKontorovich said that if a human had done it, it would be "an instant Fields Medal, no questions asked." The crypto world noticed. On Wednesday, Ethereum Foundation researcher @drakefjustin asked the industry to "calmly begin planning for bunker mode": preparing for the elliptic-curve signatures protecting most Bitcoin and Ethereum wallets to be broken by ordinary mathematics before quantum computers can do it. His worst case: "months not years." @VitalikButerin took the concern seriously but cautioned: "don't rush anything." He advised favoring hash-based schemes where possible, adding: "I personally have lost more money in botched migrations than I have lost in all hacks combined." In three bullets, the central idea: ⦿ Your wallet has a private key you keep secret and a corresponding public key ⦿ A signature made with the private key authorizes a payment; anyone can check it against the public key ⦿ The security rests on one-way math: easy to go forward, believed practically impossible to reverse on today's computers, so nobody can realistically work out your private key from your public key Nobody has proved that last step cannot be reversed; it has resisted every known attempt on today's computers. The worry is that AI might find a shortcut. There is a precedent for mathematical shortcuts: decades of work made factoring large numbers far easier than the obvious method suggested. Vitalik asks whether elliptic curves, and the newer lattice-based schemes built to resist quantum computers, hide shortcuts humans have missed but machines might find. Not everyone is convinced. Coinbase's head of cryptography, @LindellYehuda, says there is "no evidence whatsoever" that the assumptions behind elliptic curves have been weakened, and called the warning FUD. Nobody claims a breakthrough has happened; the dispute is over how seriously to prepare for one. However you weigh the risks, there are two clocks running. One is a quantum computer powerful enough to break this cryptography. It doesn't exist yet, and its development involves engineering milestones we can track. The other is a hypothetical mathematical breakthrough, which could arrive without notice and run on machines already available. Nobody can confidently date either. What would be at risk if an attacker could recover private keys from public keys? The most immediately exposed funds would be those controlled by keys that are already public. On Bitcoin, many address types hide the public key until you spend, and good wallets send change to a fresh address. On Ethereum, people commonly keep using one address, and its public key can be recovered as soon as it signs a transaction or a message. Bitcoin also has exposed keys in older outputs, Taproot outputs and reused addresses. That distinction explains Drake's advice to move funds to fresh addresses that keep their public keys hidden behind a hash. A new address is not automatically safe, though: Taproot outputs, for example, include a public key from the start. Sensible preparation starts with the ability to change. A blockchain may need to swap how it checks ownership as confidence in existing methods shifts, ideally without a network-wide upgrade. That is what crypto-agility means: engineering for flexibility. On CKB, ownership checks are programs, so new ones can be added without changing the network's rules. A signature scheme built from hash functions, the family Vitalik advises favoring where possible, is already live on mainnet: SPHINCS+. Our Chief Architect, Jan Xie, covers the quantum timeline and the challenges of migration in notes from a recent talk: https://jan.la/posts/pqc-migration-talk-notes/