OpenAI solves 722 math problems overnight, Ethereum Foundation core researcher calls for industry to enter "fortress mode"

October 8 — Ethereum Foundation core researcher Justin Drake published a post this morning calling on the blockchain industry to enter "bunker mode": gradually migrate assets to new addresses and hide public keys behind hashes to address the risk of rapid ECDSA private key recovery that AI may bring.

This warning stems from OpenAI's release of 722 mathematics papers on October 6, which included counterproofs to several long-standing conjectures (such as the Erdős unit distance problem and the n log(n) integer multiplication lower bound), showing that mathematical superintelligence is rapidly approaching.

Drake suggested that large holders and trading platforms take the lead and avoid panic; he also noted that smaller BTC wallets still have some protection due to "Satoshi's shield"; and called for accelerating the adoption of hash-based cryptography (such as SPHINCS) as the standard for the post-AI era.

In response to Drake's call, Vitalik suggested that, when operationally convenient, funds should preferentially be stored in addresses that have not yet initiated transactions, but he did not recommend that users rush to migrate assets immediately, because migration itself also carries the risk of fund loss. He also suggested that multisig wallets use off-chain signature confirmation as much as possible to avoid publicly exposing signature information.