Bitcoin and ether holders urged to enter ‘bunker mode’ against possible AI attacks

What it is

Justin Drake (Ethereum researcher) and Vitalik Buterin's "bunker mode" warning, posted on X on October 7, 2026, that AI math progress could break elliptic-curve wallet signatures on Bitcoin and Ethereum within months. The warning is timed to OpenAI's October 6 release of 722 math manuscripts from an unreleased model, including a new limit on fast matrix multiplication (a problem open since 1969) that an outside researcher independently reverified within a day.

How it works

Elliptic-curve cryptography relies on structured mathematical patterns to make deriving a private key from a public key impractical; Drake argues these "tidy patterns" are exactly what a sufficiently advanced AI could learn to exploit, unlike hash functions, which are designed to minimize exploitable structure. This is a classical (non-quantum) break risk — no quantum hardware required — which is why Drake's worst case ("months, not years") lands far earlier than the Ethereum Foundation's quantum-resistance migration target.

Preconditions / Affected

• Bitcoin addresses whose public keys are already exposed onchain (millions of BTC, per prior CoinDesk reporting).
• Any Ethereum account that has ever sent a transaction, since that reveals its public key — this also covers stablecoins and tokenized funds issued on Ethereum.
• Lattice-based post-quantum schemes (the basis of NIST's approved signature standard) are separately flagged by Buterin as potentially weakened by the same AI math trend, drawing an analogy to how GNFS improved on naive factoring.

What to do

• Drake: large/sophisticated holders should gradually migrate funds to new addresses whose public keys have never appeared onchain, removing the exposed starting point an attack would need.
• Buterin: agrees public-key exposure should be reduced where practical, but warns against rushing — "I personally have lost more money in botched migrations than I have lost in all hacks combined."
• Ethereum's long-term roadmap is shifting toward hash-based signatures over lattice-based ones, given Buterin's concern about lattice security eroding under AI-driven math advances over the next two years.

Caveats

No practical attack on Bitcoin or Ethereum wallet keys has been demonstrated, and CoinDesk found none in the research it reviewed — this is a preparedness argument, not a confirmed exploit. The timeline is explicitly a "worst case" estimate from Drake, contrasted against the Ethereum Foundation's own December 2029 quantum-resistance target. The cited real-world incidents (Coldcard firmware bug, BTCPay/Lightning thefts, Core Lightning bugs, Anthropic's DeFi exploit demo) establish AI-assisted vulnerability discovery as already real, but none of them are breaks of elliptic-curve signatures themselves.