Justin Drake: AI Risks Breaking Bitcoin ECDSA

Justin Drake of the Ethereum Foundation warned on Oct 7 that AI could break ECDSA signature security for Bitcoin and Ethereum wallets in the worst case within months.

From Quantum Fears to AI Math

The alert follows a March Google Quantum AI paper co-authored by Drake estimating fewer than 500,000 physical qubits needed to attack the cryptography, down 20x from prior estimates. Drake assigns at least 10% odds of a quantum computer recovering a private key by 2032. The latest post shifts focus to OpenAI's unreleased model releasing 722 math papers on Oct 6, raising whether AI will discover shortcuts in elliptic curve math similar to decades of RSA improvements.

Exposure and Practical Limits

Only addresses that have spent reveal their public key on-chain. An Ark Invest and Unchained March paper pegged 34.6% of Bitcoin supply, or 6.9M BTC, as exposed via reused addresses, old P2PK outputs and Taproot. Drake defines a break as recovering one private key in roughly one week on a large GPU cluster. Coinbase cryptography head Yehuda Lindell states there is no evidence elliptic curves are close to failing.

Bitcoin price action on the 4h chart shows BTC trading at $82410.47 inside the Bollinger Bands with upper resistance at $87184.35 and lower support at $81846.82. The structure remains bearish as price sits below the EMA50 at $84388.59 while the EMA200 at $82180.22 acts as immediate floor. Momentum indicators confirm pressure with RSI(14) at 28.23 in oversold territory and MACD at -680.22 printing a death cross, suggesting any bounce toward the 50-EMA may stall before continuation lower.

Traders monitoring BTC price prediction models and AI industry impact on cryptography should note that no published OpenAI paper targets ECDSA and that hash functions used in mining remain untouched. Holders reduce risk by moving coins from addresses that have already spent.