Ethereum co-founder Vitalik Buterin said on October 7 that it is "plausible" artificial intelligence could break lattice-based cryptography within the next two years, urging the crypto industry to minimize exposure to AI-vulnerable cryptography in addition to quantum-vulnerable schemes.

His warning landed one day after Ethereum Foundation researcher Justin Drake called on the industry to plan for "bunker mode" — moving funds to addresses whose public keys have not yet been exposed. Drake argued that AI-accelerated advances in mathematics could break encryption before quantum computers become powerful enough to do so, the milestone commonly known as Q-Day.

"There is a good chance that the concrete security of lattices will take serious hits from the next two years of AI math," Buterin wrote on X. "If AI will bring us 50 years of math in two years, then that 50 years of math may very plausibly include a [...] level of improvement to our ability to break lattices."

Why does this complicate quantum-proofing plans?

Lattice-based cryptography is one of the leading candidates for making blockchains resistant to quantum computers, and some projects — Cardano among them — have considered adopting it. Buterin's comments expose an operational tension: users migrating to lattice-based addresses to hedge against a long-term quantum threat could land in systems vulnerable to a nearer-term AI risk.

"Potential breaks in cryptography by quantum computers were generally thought to be contained to Elliptic curve cryptography and RSA," Isabel Foxen Duke, head of Mara Foundation and co-author of BIP-360, told Decrypt. "Breaks in cryptography by classical methods (e.g. those potentially created by AI) could affect types of cryptography that are broadly considered quantum secure (e.g. lattice-based cryptography). If we include concerns about classical breaks in the mix, we have to be even more conservative."

Buterin's practical recommendation follows directly from that conservatism. Any lattice-based scheme adopted should significantly increase its lattice sizes, he said. And on hash-based alternatives, which he now views as the better route to quantum-proofing, he conceded that AI-driven mathematics could theoretically break hashes as well — leaving no fully unassailable option.

Should users move their funds now?

Buterin says no. "I don't recommend anyone scramble to move their funds to new wallets today," he posted. "But we should take the risks to cryptography from AI-accelerated math seriously, and minimize our exposure to not just quantum-vulnerable cryptography, but also potentially AI-vulnerable cryptography."

The cheapest defensive measure he proposed is behavioral rather than cryptographic: keep funds in addresses that have never made a transaction. Such addresses have not exposed their public keys, which significantly reduces the attack surface.

"I wouldn't treat Vitalik's two-year scenario as a deadline. The post doesn't demonstrate that these systems have been broken," Ananda Banerjee, founder of analytics firm Charlie Quant Lab, told Decrypt. "I would prepare carefully rather than rush to move funds."

What changed in a month?

The warning marks a shift in Buterin's posture. Just a month earlier, he argued that AI would not doom crypto security because defensive capabilities would keep pace with offensive ones. His latest comments suggest the industry's quantum-migration planning may have been pointing in slightly the wrong direction — prioritizing lattice schemes that could face an earlier classical break than the quantum threat they were designed to withstand.

For protocol teams, the immediate consequence is a design constraint rather than an emergency: any post-quantum migration roadmap now has to price in the possibility that lattice parameters sized against quantum adversaries may be undersized against AI-accelerated classical cryptanalysis, and hash-based fallbacks are no longer a guaranteed safe harbor. How Ethereum's wallet ecosystem and competing chains reconcile those two threat models will shape post-quantum migration timelines over the coming two years.