Vitalik Buterin urges calm as AI raises new fears over Bitcoin and Ethereum cryptography security

Vitalik Buterin, co-founder of Ethereum, has cautioned that artificial intelligence-driven breakthroughs in mathematics could potentially undermine both current crypto signatures and select post-quantum security measures.

On October 7, Buterin suggested that cryptocurrency owners minimize their exposure where feasible, yet he strongly advised against rushing to move funds into fresh wallets. He emphasized that errors made during migration represent an immediate hazard far greater than any currently proven cryptographic breach.

He posted the following on X:

“I don’t recommend anyone scramble to move their funds to new wallets today. But we should take the risks to cryptography from AI-accelerated math seriously.”

OpenAI’s math advances trigger warnings

His remarks followed a recommendation by Ethereum researcher Justin Drake, who urged the broader industry to start preparing for “bunker mode.” This suggestion included transferring assets to brand-new addresses whose public keys have never been disclosed onchain.

Drake pointed out the possibility that advancements in artificial intelligence might speed up mathematical attacks targeting the elliptic-curve cryptography utilized by both Bitcoin and Ethereum.

These discussions escalated after OpenAI shared new mathematical findings produced by an internal frontier model on October 6, which included machine-generated proofs formalized in Lean. OpenAI did not report any attacks against blockchain cryptography, and neither Buterin nor Drake presented evidence that ECDSA has actually been practically cracked.

However, Buterin’s caution reaches past that short-term concern. He specifically highlighted lattice-based cryptography—such as ML-DSA—as a sector where AI-powered mathematical breakthroughs could diminish assumed security margins over the upcoming two years.

This introduces a tougher challenge for a sector already getting ready for quantum computing. Lattice-based architectures stand out as primary candidates created to substitute for cryptographic systems vulnerable to sufficiently capable quantum computers. In 2024, NIST standardized ML-DSA to serve as a post-quantum digital-signature algorithm.

Buterin maintained that the underlying mathematical foundations supporting such frameworks could harbor vulnerabilities that researchers have yet to uncover.

He likened this threat to the evolution of integer factorization. Enhancements like the general number field sieve drastically cut down the computational work needed to challenge RSA compared to basic naive approaches, which ultimately compelled cryptographic systems to adopt much larger keys.

According to Buterin, artificial intelligence could compress decades of equivalent mathematical development into a significantly tighter timeframe, potentially unlocking comparable breakthroughs against elliptic curves or lattice problems.

“If AI will bring us 50 years of math in 2 years,” Buterin wrote, those advancements might encompass breakthroughs that significantly enhance attacks directed at lattice cryptography.

Ethereum leans further into hash-based security

That worry helps clarify why Ethereum developers are increasingly leaning toward cryptographic frameworks centered primarily around hash functions.

Buterin noted that Ethereum’s “lean” development trajectory has shifted toward hash-only mechanisms throughout the past year, steering clear of lattice-based signatures like ML-DSA and Falcon while minimizing dependence on lattice-based commitments inside zero-knowledge proofs.

Hash-based signature systems—including WOTS and SPHINCS—take on a more prominent role within that strategy because they rely far less on mathematical designs that might provide attackers with unexpected shortcuts.

Buterin argued that structured mathematical elements create greater room for AI-assisted discoveries. For instance, elliptic curves rely on specific attributes like group operations and other exploitable mathematical connections. Similarly, lattice systems depend on families of complex problems whose practical resistance could shift if researchers discover more streamlined attacks.

Hashes are purposefully built to reveal considerably less structure.

This distinction does not resolve every challenge. Public-key encryption—which secures messaging apps, anonymous communication networks, websites, and various other platforms—cannot typically be constructed using hashes alone. Those specific applications demand mathematical structures that establish a trapdoor, permitting authorized individuals to decrypt data.

Buterin proposed that architectures depending on those specific structures might eventually require significantly larger security parameters. He stated that multiplying key sizes by as much as a factor of ten could become reasonable if AI drastically enhances cryptanalytic techniques.

The resulting implications would reach far beyond blockchains. Secure messaging platforms, virtual private networks (VPNs), Tor, encrypted web traffic, and privacy protocols could encounter comparable compromises between performance and expanded security buffers.

Buterin warns against panic migrations

For cryptocurrency users, the immediate precautionary steps remain relatively narrow.

Buterin noted that individuals who can easily maintain funds in addresses that have never executed a transaction might secure extra protection, as their public keys stay concealed behind address hashes. The moment an account broadcasts a transaction, its public key becomes visible.

Furthermore, multisig operators can diminish their exposure by gathering signatures offchain and cycling signer keys following operations, thereby restricting the duration an exposed key stays useful if ECDSA security weakens.

Even so, Buterin cautioned against treating these preventative measures as an emergency directive to migrate assets. He shared that he has personally lost greater sums of money through unsuccessful wallet migrations than through security hacks.

Consequently, wallet providers, custodians, and protocol developers face a far more complex assignment than simply rotating keys. They are required to prepare for rapidly evolving cryptographic dangers while steering clear of upgrades that might trigger operational errors of their own.

?Frequently Asked Questions

01What are the main risks AI poses to cryptocurrency cryptography?

AI-driven advances in mathematics could potentially accelerate attacks on current elliptic-curve cryptography used by Bitcoin and Ethereum, as well as weaken assumed security margins in post-quantum defenses like lattice-based systems.

02Should I move my crypto assets to new wallets right away?

No. Vitalik Buterin advises against panic migrations, noting that human errors during wallet transfers present an immediate danger greater than any currently proven cryptographic break.

03What is Ethereum’s developer strategy regarding post-quantum security?

Ethereum developers are increasingly favoring hash-based cryptographic designs—such as WOTS and SPHINCS—while avoiding lattice-based signatures like ML-DSA and Falcon to reduce vulnerability to unexpected mathematical shortcuts.