Ethereum Researcher Warns AI Could Break Crypto Signatures
Justin Drake, a researcher at the Ethereum Foundation, has issued a warning that AI-assisted mathematics could compromise the signature schemes used by Bitcoin and Ethereum wallets before quantum computers pose a threat.
Drake shared his concerns on X, and urged the industry to prepare for what he called “bunker mode.” This warning is significant because it suggests a much shorter timeline for cryptographic threats, potentially shifting from a quantum horizon measured in years to a possibility Drake frames as months, not years in a worst-case scenario.
The Researcher’s Claims
Drake’s primary concern revolves around the Elliptic Curve Digital Signature Algorithm (ECDSA), which Bitcoin and Ethereum utilize to verify that transactions originate from the correct private key. When a wallet signs a transaction, its public key becomes visible on the blockchain. If ECDSA were compromised, an attacker could potentially reverse-engineer the private key from the exposed public key, thereby gaining control of the wallet’s funds.
To mitigate this risk, Drake recommends a controlled migration of funds to new addresses where public keys remain hidden behind a hash, along with key rotation after each signing event. He advised holders against panic or hasty actions, noting that moving assets doesn’t require new wallets or cryptographic methods. He specifically mentioned institutions like Binance, Bitbank, Robinhood, Bitfinex, and Tether as needing to strengthen their cold storage security.
Drake also highlighted a recent surge in AI-driven mathematical advancements, including 722 results published by OpenAI, as evidence that long-held assumptions are becoming less robust. He specifically stated that “Elliptic curves feel especially vulnerable to superintelligence,” arguing that hash functions are inherently more resistant to the structured attacks that elliptic curves might invite.
How AI Alters the Pre-Quantum Threat
Drake outlined the following points regarding the potential AI threat:
- Signature Scheme at Risk: ECDSA, which is widely used across Bitcoin and Ethereum wallets.
- Definition of “Break”: The ability to recover a private key within approximately a week using hardware comparable to a large GPU cluster.
- Worst-Case Timeline: Months, not years, preceding a potential “Q-Day” (when quantum computers pose a threat).
- Partial Protection: Wallets holding under 50 BTC are somewhat shielded by what Drake termed “Satoshi’s shield,” which is associated with roughly 20,000 exposed addresses each containing 50 BTC.
- Proposed Solution: A faster transition to hash-based signatures, such as SPHINCS, aligning with NIST‘s SLH-DSA, FIPS 205.
Drake’s assertion is a projection, not a report of an actual breach. There is currently no evidence that AI has successfully recovered a live wallet key, and the OpenAI math results he referenced do not involve cryptanalysis of secp256k1. His timeline represents a worst-case estimate from a public statement, not a definitive measurement.
Nevertheless, the warning carries significant weight given its source. Earlier this year, Drake estimated a 10% or greater chance of Q-Day by 2032, and the Ethereum Foundation already maintains a dedicated post-quantum team. This introduces a second critical timeline for custodians and wallet developers to monitor. Vendors like BitGo have begun assessing Bitcoin’s quantum resistance, and identity systems such as the TIP Protocol, currently in testing with simulated data, are proposing signature-based trust models that would face similar security questions.
For more information on the quantum threat to crypto signatures and NIST’s post-quantum standards, refer to QNews coverage on Q-Day preparation and hash-based cryptography.
Learn more