Ethereum co-founder Vitalik Buterin has raised concerns that rapid advancements in artificial intelligence (AI) may compromise the security of certain cryptographic systems within the next two years. His comments came in response to a suggestion from Ethereum Foundation researcher Justin Drake, who called on the industry to adopt a heightened defensive posture amid growing AI-driven risks. Buterin emphasized that the threat extends beyond quantum computing, traditionally considered the main future challenge for cryptography.
Concerns over lattice-based cryptography and AI advancements
Buterin highlighted that AI’s progress in complex mathematical fields, such as number theory and algebra, could uncover weaknesses in lattice-based cryptography systems. These include ML-DSA and fully homomorphic encryption (FHE), which are designed to provide security amidst emerging computational threats. He explained that Ethereum is also investigating alternative cryptographic options, such as hash-based signature schemes like WOTS and SPHINCS+, which are less dependent on intricate mathematical assumptions.
Mini dictionary: Fully homomorphic encryption (FHE) allows computations to be performed directly on encrypted data, enabling data privacy without needing to decrypt the inputs first. WOTS (Winternitz one-time signature) and SPHINCS+ are types of hash-based digital signature schemes considered resistant to both quantum and some AI-driven attacks.
User reactions and security recommendations
Amid ongoing debate within the crypto community, Buterin urged users not to panic over potential AI threats. He advised against rushing to migrate wallets or making abrupt changes to security practices, warning that poorly planned transitions could result in greater losses than the risks posed by AI-driven attacks themselves. Buterin shared a personal anecdote, stating, “I personally have lost more money in botched migrations than I have lost in all hacks combined.” Instead, he recommended that users adopt fresh, unused wallet addresses where public keys have not been previously exposed, use larger key sizes for certain encryption systems, and consider offline or off-chain confirmations for multisignature wallets.
I personally have lost more money in botched migrations than I have lost in all hacks combined. Use fresh addresses and be cautious with security changes.
Several X (formerly Twitter) users contributed to the discussion. One user, @walicoin, focused on the “extra algebraic structure in Ring/Module Learning With Errors (LWE),” referencing prior vulnerabilities in cryptographic schemes including a 2022 classical attack on SIKE and issues in cyclotomic ideal SVP. The user argued that hash-based schemes may bypass these vulnerabilities, although current assurances remain heuristic.
Another participant, @the_smart_ape, framed crypto security as an ongoing mathematical puzzle, rather than an unsolvable problem. He cautioned that AI could accelerate mathematical discovery to the point where “quantum-safe might not be AI-safe.” He encouraged users to maintain disciplined security practices instead of reacting hastily.
X user @stacymuur broadened the perspective, noting that a successful quantum attack on modern cryptography would undermine not only crypto but also bank accounts, brokerage accounts, and major online platforms. The user emphasized that such an event would threaten the very foundation of the global financial system and internet security.
If quantum technology one day breaks modern cryptography, the consequences would extend beyond crypto to bank, brokerage, Google, and Apple accounts—crypto would be the least of the problems, and this would signal the endgame for the internet.
Blockchain migration and infrastructure challenges
The conversation also addressed the feasibility of migrating existing blockchain systems to stronger cryptographic standards. Such a migration generally requires updating the cryptographic process for transaction authorization, not physically transferring all tokens or assets to new wallets. For instance, Ethereum developers could upgrade current accounts to support safer signature schemes, which would protect existing ETH, tokens, and decentralized finance positions without mass asset transfers.
Ethereum, an open-source blockchain network and the second largest cryptocurrency platform, is pursuing a post-quantum roadmap centered on “signature agility.” This approach allows individual accounts to adopt new authentication methods independently, preventing the need for coordinated, network-wide action. Ethereum is currently exploring EIP-8141 as part of this effort, with the goal of securing core infrastructure by approximately 2029.
Security upgrades not only involve account signatures but also critical elements such as Boneh–Lynn–Shacham (BLS) validator signatures, Kate–Zaverucha–Goldberg commitments, and zero-knowledge systems. BLS signatures are compact at roughly 96 bytes, while potential replacements like hash-based leanXMSS signatures are estimated at about 3,000 bytes, presenting significant technical challenges for system efficiency and scalability.
| Signature Scheme | Estimated Size (Bytes) | Adoption Status |
|---|---|---|
| BLS Signature | 96 | Active (Ethereum validators) |
| leanXMSS (hash-based) | 3,000 | Potential replacement (in research) |
By contrast, Bitcoin faces a complex governance challenge: a protocol-level migration could impact coins whose owners are inactive or have lost access, complicating any network-wide security overhaul.
