Ethereum co-founder Vitalik Buterin has responded skeptically to forecasts suggesting that artificial intelligence (AI) might lead to a 50% decline in Bitcoin’s value within the next two years.
This discussion was sparked by Silicon Valley investor and AI risk analyst Liron Shapiro, who indicated a 50% chance that Bitcoin could see a fall greater than 50% due to AI potentially eroding investor confidence in the network’s security and reliability.
In contrast, Buterin presented a different viewpoint, stating his perspective:
My fundamental view is that I am optimistic about long-term cybersecurity and believe the main challenge lies in facilitating the transition.
Buterin believes that Bitcoin can effectively manage security issues that do not depend on collective decision-making. This includes attacks that might compromise clients, mining pools, and other infrastructure. He expressed minimal concern about AI fundamentally disrupting Bitcoin’s hashing algorithms or proof-of-work mechanism.
While typically he might challenge Shapiro to a bet on this outcome, Buterin noted that his existing cryptocurrency investments already represent a substantial wager, comprising nearly 90% of his wealth. He further implied that the same rationale could apply to Ethereum.
This disagreement highlights an increasing divide regarding the implications of advancing AI for crypto security.
Shapiro’s argument does not necessitate AI to breach Bitcoin’s cryptographic foundations. A series of attacks that reveal previously assumed secure weaknesses could undermine investor confidence, even if those vulnerabilities are eventually patched by developers.
Buterin’s perspective differs in that he anticipates most security issues driven by AI to be resolvable. He believes defenders will ultimately leverage the same technological advancements that attackers utilize.
AI threats are already putting crypto defenses to the test
Evidence throughout the cryptocurrency sector indicates that the threats highlighted by Shapiro are turning into reality, even without a breakthrough in Bitcoin’s core cryptography.
In August, Boltz, a Bitcoin swap provider, had to indefinitely halt its operations after months of AI-assisted probing resulted in several contained exploits, surpassing the capability of its small development team to address them in time.
Although Boltz’s non-custodial model safeguarded customer assets from loss, the company faced financial setbacks due to these attacks, leading to the conclusion that it could no longer safely maintain the service.
This incident highlighted how AI might exert pressure on the broader infrastructure of Bitcoin without compromising its proof-of-work or breaking its cryptographic foundations. Instead, attackers utilized automation to rapidly identify weaknesses, outpacing the defenders’ ability to investigate, patch, and implement solutions.
This kind of threat is anticipated to become more prevalent as offensive AI capabilities become more sophisticated.
Deddy David, CEO of blockchain security firm Cyvers, previously remarked to CryptoSlate that the financial risks stemming from AI-enhanced crypto attacks could potentially reach hundreds of millions or even billions of dollars.
“If AI can pinpoint vulnerabilities across the core internet infrastructure, the cryptocurrency market will likely feel the impact first,” David stated.
The risk is widespread across various components, including wallets, bridges, exchanges, smart contracts, and blockchain network software, providing significantly more entry points for attackers than Bitcoin’s consensus mechanism.
This distinction lies at the heart of the debate between Buterin and Shapiro, as AI doesn’t need to compromise SHA-256 to inflict substantial damage within the Bitcoin ecosystem.
The unresolved question is whether these attacks will remain manageable through developer updates and enhanced protections, as Buterin believes, or escalate to a level that undermines the security assumptions that investors have integrated into Bitcoin.
The tech industry is racing to bolster defenses
The broader tech industry is increasingly resembling an arms race, striving to ensure that defensive AI evolves at least as quickly as offensive capabilities.
Anthropic has limited public access to its Claude Mythos model due to its potential to autonomously detect and exploit software vulnerabilities.
Instead, the company is focusing those capabilities on defenders through Project Glasswing, a collaboration involving tech giants such as Amazon Web Services, Google, Microsoft, and JPMorgan Chase.
This initiative employs Mythos Preview to identify and resolve vulnerabilities in critical systems ahead of malevolent actors equipped with similar technologies.
Anthropic has committed up to $100 million in usage credits toward this effort.
The movement to enhance defenses has expanded beyond individual entities.
Over 100 organizations, including Google, Microsoft, Anthropic, and OpenAI, have signed an open letter urging governments and companies to recognize the likelihood of AI-driven cyberattacks becoming increasingly frequent and sophisticated in the near future.
Financial institutions, payment companies, and major technology players—including Capital One, Mastercard, Visa, Adobe, Oracle, and IBM—are also backing the call.
The letter emphasized that current cybersecurity methods will likely fall short as AI systems become more advanced, particularly within historically underfunded critical infrastructure sectors.
Signers advocated for both governments and tech firms to supply advanced defensive AI and security testing to entities such as hospitals and water utilities.
Cryptocurrency is part of the same race but faces a unique challenge: many of its systems manage instantly transferable financial assets and function continuously on publicly accessible infrastructure.
This complexity makes resolving the dispute between Buterin and Shapiro a difficult task, as it’s not solely about whether AI can “break Bitcoin.” The pressing question is whether attackers will hold a definitive edge during the interim before defensive technologies, auditing practices, and network structures can catch up.