Experts Assess the Risk of Hacking Cryptocurrency Wallets Using AI

cryptonews.ruPublished on 2026-10-12Last updated on 2026-10-12

Abstract

Cybersecurity and blockchain industry experts have assessed the risk of cryptocurrency wallet hacks due to potential breakthroughs in mathematical research using artificial intelligence. The discussion intensified after Ethereum Foundation researcher Justin Drake stated that AI could threaten cryptographic security sooner than quantum computers and urged investors to prepare for a "bunker mode." Vitalik Buterin agreed on the necessity to account for such risks while also calling for calm. The editorial team at Incrypted consulted experts on the realism of this scenario and how the blockchain industry can prepare for it.

Cybersecurity and blockchain industry experts have assessed the risk of hacking cryptocurrency wallets due to possible breakthroughs in mathematical research using artificial intelligence. The discussion intensified after Ethereum Foundation researcher Justin Drake stated that AI could compromise cryptographic security sooner than quantum computers and urged investors to prepare for a 'bunker mode'.

Vitalik Buterin agreed on the need to consider such risks while urging against panic. The Incrypted editorial team asked experts how realistic such a scenario is and how the blockchain industry can prepare for it.

Related Questions

QWhat is the main concern raised by Ethereum Foundation researcher Justin Drake regarding AI and cryptography?

AEthereum Foundation researcher Justin Drake warned that AI could potentially break cryptographic protections safeguarding cryptocurrency wallets earlier than quantum computers, urging investors to prepare for a 'bunker mode'.

QWhat was the reaction of Vitalik Buterin to the risks posed by AI to cryptographic security?

AVitalik Buterin acknowledged the need to consider such risks but simultaneously urged against panic.

QWho was asked for their opinion on the feasibility of an AI breakthrough threatening crypto wallets?

AThe editorial team of Incrypted asked cybersecurity and blockchain industry experts for their opinion on how realistic such a scenario is.

QWhat specific threat was compared to the potential risk from AI in the article?

AThe article compares the potential threat from AI to the long-anticipated threat from quantum computers to cryptographic security.

QWhat did the experts discuss besides assessing the risk of AI hacking crypto wallets?

AExperts also discussed how the blockchain industry could potentially prepare for such a scenario involving AI breakthroughs in mathematical research.

Related Reads

Ethereum Plans to Move Post-Quantum and Anti-AI Upgrades Forward to 2027: Ethlabs Proposes Restructuring the Hegotá Hard Fork Plan

Ethlabs, a non-profit Ethereum R&D lab, has submitted the pqai27 community proposal to the Ethereum community forum, calling for 2027 to be the target year for achieving quantum resistance and AI resistance on Ethereum. With Ethereum's on-chain TVL exceeding $41.16 billion as of November 12, the proposal argues that Ethereum should leverage its existing hash-based cryptography and formal verification foundations to reinforce its position as the global value settlement layer as AI evolves rapidly and quantum timelines accelerate. The direct motivation is the compression of external timelines: the U.S. government has moved its quantum-resistance target from 2035 to 2030, and Google plans to complete its full cryptographic migration by 2029. AI attack capabilities are also advancing—OpenAI has used frontier models to find and patch roughly 1,000 Linux vulnerabilities and solve hundreds of math problems. "AI resistance" means relying on minimal hash primitives, formal verification, and client diversity to harden the base layer. The roadmap splits 2027 into two stages. The baseline "pqai-ready" stage requires quantum resistance at the consensus layer and ensures users can migrate to quantum-resistant accounts via the Frames account abstraction proposal EIP-8141; if existing cryptography fails early, the network can degrade to calldata to keep running. The advanced "pqai-scaled" stage targets full-capacity operation, introducing EIP-8288 for quantum-resistant signature aggregation and EIP-8479 for quantum-resistant data availability, so quantum-resistant transactions and blobs can regain normal throughput after scaling. To this end, Ethlabs recommends restructuring the scope of the Hegotá hard fork, stripping out non-critical upgrades and reassessing the impact of the FOCIL anti-censorship proposal (EIP-7805) on progress and compatibility; if the execution layer is decoupled from the consensus layer, Frames could ship independently first. The consensus layer must choose between retrofitting the existing Gasper implementation for quantum resistance and moving to decoupled consensus. Ethlabs announced it is withdrawing its push for Quick Slots (EIP-8198) in Hegotá, deferring it to 2028, while Glamsterdam and its scaling plans proceed as normal. The roadmap remains an Ethlabs community initiative; any Hegotá fork restructuring or consensus-layer switch requires further discussion and agreement among Ethereum's All Core Devs. Ethlabs notes that AI is also accelerating formal verification and zero-knowledge proof benchmark optimization, and that advancing pqai27 aims to consolidate Ethereum and ETH's role as the trust root for global value settlement during extreme security cycles.

marsbit12m ago

Ethereum Plans to Move Post-Quantum and Anti-AI Upgrades Forward to 2027: Ethlabs Proposes Restructuring the Hegotá Hard Fork Plan

marsbit12m ago

Former OpenAI Researcher Changes His Tune: Super AI Researcher Could Arrive as Early as Next Year

Former OpenAI researcher Daniel Kokotajlo and the AI Futures Project have revised their timelines, now predicting superhuman AI researchers (SARs) could arrive as early as 2027 rather than 2029. The key driver is a METR assessment of Anthropic, which found that since October 2025, the benchmark score growth rate of Anthropic's internal models reached about 1.9x the previous pace (95% CI: 1.5–2.8x) — roughly two years' worth of progress in one year. METR estimates that by October 2026, about 80% of the extra acceleration may come from AI itself, though this is a subjective guess with high uncertainty. AI is already helping researchers write experiment code, run tests, and try out more ideas. Using METR's estimates, the team calculated that by early November 2026, AI-driven acceleration alone could push overall model progress to 1.7x the historical rate — a speed that, in the original AI 2027 scenario, wouldn't be reached until around April 2027, about five months later. Notably, METR says most of the evidence underpinning its core conclusions was redacted from the public report at Anthropic's request, to avoid helping competitors; only the benchmark curves are publicly verifiable. Depending on the method used, SAR arrival estimates range from February 2027 to mid-2028. If acceleration compounds across model generations, superintelligence could arrive sooner than expected.

marsbit32m ago

Former OpenAI Researcher Changes His Tune: Super AI Researcher Could Arrive as Early as Next Year

marsbit32m ago

Aave V3.8 Officially Launches: Covering 15 Public Chains, $15.2 Billion Liquidity Pool, Introducing Low-Gas-Cost Reentrancy Protection

Aave has launched V3.8 across 15 active V3 markets, following approval by the Aave DAO and development by Aave Labs. According to DeFiLlama, Aave's total value locked exceeds $16.4 billion, with active V3 markets holding over $15.23 billion. The upgrade was deployed via AIP #541, completing a 22-day governance process that began with a forum proposal on October 21 and a Snapshot vote on October 27. V3.8 covers active instances on Ethereum (Core and Lido), Plasma, Avalanche, BNB Chain, Monad, Base, and Arbitrum, along with Mantle, X Layer, Polygon, Linea, Optimism, Gnosis, MegaETH, and Celo. Legacy markets that were frozen with zero caps were deliberately skipped. The release introduces kernel-level reentrancy protection for external callback operations on liquidity pools. Traditionally, preventing cross-function reentrancy requires writing a lock flag before state changes, but after EIP-3529 capped SSTORE refunds at 20%, maintaining persistent locks in high-frequency lending pools became gas-intensive. V3.8 instead uses EIP-1153 transient storage, where TSTORE and TLOAD cost only 100 gas per read or write and automatically clear at transaction end, providing reentrancy hardening at minimal cost. Aave Labs addressed developer concerns that the lock might block liquidation composability within flash loan callbacks. The flashLoan and flashLoanSimple entry points only check the lock status without setting it, creating asymmetric isolation: users can still compose liquidations, deposits, or repayments inside a flash loan callback, but attempts to initiate a flash loan from within a regular pool operation callback are rejected and reverted. Security verification included Certora formal verification and independent external audits before the AIP vote. The new contract code is published in the official Aave V3 repository under the Business Source License. Developers also noted that downstream integrators reading view functions such as account health factors or reserve data during callbacks should account for intermediate ledger states and ensure peripheral risk controls remain compatible.

marsbit33m ago

Aave V3.8 Officially Launches: Covering 15 Public Chains, $15.2 Billion Liquidity Pool, Introducing Low-Gas-Cost Reentrancy Protection

marsbit33m ago

Trading

Spot
活动图片