Vitalik Buterin Backs "Bunker Mode" for Crypto Over AI Progress in Math

The crypto industry has been told to prepare for the worst-case scenario. Justin Drake, a researcher at the Ethereum Foundation, proposes going into "bunker mode": he estimates that AI could undermine wallet security within months. According to Cointelegraph, Vitalik Buterin backed this position. The headline is loud, but the available material does not describe a specific attack, and that is worth keeping in mind.

What is known and what is not

The original source supports three points. The warning comes from Drake, a prominent researcher in the Ethereum ecosystem. He names a short horizon: months. And the trigger is rapid AI progress in mathematics. Our source contains no hacking mechanism, no model name and no verifiable results. That makes this an expert risk assessment, not a confirmed incident. So far, not a single compromised wallet figures in this story.

The Hacker News discussion is small: 60 points and 49 comments. The community is clearly interested, but there is no consensus.

Why math, specifically

Wallet security rests on mathematical problems that are considered practically unsolvable. Bitcoin and Ethereum use elliptic-curve signatures (ECDSA on secp256k1, plus Schnorr in Bitcoin's case). A private key can be derived from a public one only in theory; in practice it takes prohibitively large computation. The industry's classic fear is quantum computers and Shor's algorithm. Here the risk comes from a different source: models that get better at proving theorems and finding unconventional solutions could speed up the search for weaknesses in the schemes themselves or in their implementations.

The material does not make clear which of these Drake has in mind. For risk assessment, that distinction is fundamental: a vulnerability in a specific implementation is fixed with an update, while undermining the mathematical foundation itself would require replacing the cryptography wholesale.

What "bunker mode" means

The source does not define the term, but the logic is readable: tighten security in advance, without waiting for a proven attack. For the industry, this usually means moving to post-quantum and hash-based signature schemes, rotating keys and ending address reuse. Drake has previously promoted ideas in Ethereum tied to a simplified protocol built on hash-based signatures, so his position fits a long-standing line of work.

Key facts and reference points:

• Timeline in the warning: months, not years.
• Author of the warning: Justin Drake, Ethereum Foundation. Backed by: Vitalik Buterin.
• Vulnerable area: wallet signature cryptography (elliptic curves).
• Ready-made standards exist: in August 2024, NIST published FIPS 203, 204 and 205 (ML-KEM, ML-DSA, SLH-DSA).
• Hacker News reaction: 60 points, 49 comments.

How to act without panicking

For crypto holders, a sensible minimum does not depend on how accurate the forecast is. Do not reuse an address after sending funds: in a number of schemes, the public key becomes visible only after the first spend. Keep large sums at addresses where the public key has not yet been revealed. Follow announcements from the Ethereum Foundation and Bitcoin Core developers about plans to migrate to new signatures.

For exchanges, custody providers and wallet developers, the takeaway is tougher. They need an inventory of every place where ECDSA keys are used, and a replacement plan that can be launched within weeks. If Drake's forecast turns out to be overstated, the losses will be modest: engineers' time. If it turns out to be understated, the cost of the mistake is on a different scale entirely.

For now, the main question for the authors of the warning is simple: which specific AI results in mathematics do they consider alarming, and can those results be verified independently? Without an answer, "bunker mode" remains a cautious strategy, not a response to a proven threat.