Phemex CEO警告：AI采用对加密货币产生“净负面影响”

Phemex CEO 费德里科·瓦里奥拉：AI 对加密货币行业目前仍是“净负面”影响

Phemex 首席执行官费德里科·瓦里奥拉（Federico Variola）对“人工智能（AI）将自动使加密货币受益”的观点提出反驳。他指出，迄今为止，AI 技术通过从该领域抽离流动性以及增加网络攻击的规模和复杂性，实际上起到了“净负面”作用。在接受 Cointelegraph《Chain Reaction》栏目采访时，瓦里奥拉警告称，更快的漏洞利用周期和不断上升的安全成本可能会迫使行业走向更大的中心化。

他的这番言论是在 Phemex 今年早些时候宣布 AI 转型计划之后发表的，该计划包括在产品开发内部运营中全面嵌入 AI。然而，瓦里奥拉最新的评论侧重于 AI 对加密货币安全性和去中心化更广泛的影响——特别是对于构建协议的团队以及试图安全使用 DeFi 的零售用户而言。

核心要点

• 瓦里奥拉表示，AI 已将资本从加密货币领域转移，同时也赋能了攻击者，并增加了网络安全的负担。
• 随着 AI 使漏洞发现和武器化变得更快，较小的协议团队可能难以跟上步伐。
• 由于用户面临更多设备被入侵和社会工程学攻击的途径，这些威胁可能会降低零售用户对 DeFi 的兴趣。
• 尽管攻击手段日益复杂，但安全专家也看到了 AI 在防御方面的潜力。
• 瓦里奥拉认为，AI 代理是投资者的辅助工具，而非取代人类在交易中的决策制定。

流动性流失与“中心化”风险

瓦里奥拉的批评主要集中在激励机制上。他认为，AI 的整体势头已将资源——包括资本——吸引向专注于 AI 的项目，而不是流入加密基础设施和生态系统。同时，他辩称，AI 降低了恶意活动的门槛，使不法分子能够更有效地进行协议漏洞利用和社会工程学攻击。

在他看来，净效应是一种结构性转变，可能有利于资金更雄厚的大型参与者。“很难想象一个 AI 会专门 favor 加密货币行业的世界，”瓦里奥拉说，并补充道，许多正在推进的“修复”措施反而会鼓励中心化，而不是减少它。

这种观点之所以重要，是因为去中心化不仅是加密货币的哲学目标，还取决于团队能否在不依赖昂贵且持续堆叠防御体系的情况下维持安全性。如果保持安全的成本上升速度快于行业缓解风险的能力，权力自然会集中。

现实世界的漏洞利用凸显了 AI 辅助攻击开发的节奏

Variola's concerns are not purely theoretical. He linked the comments to known instances where AI is directly or indirectly linked to exploit cycles in the realm of cryptographic security.

Earlier this year, media reported an incident in July in which attackers stole approximately $116 million in Bitcoin from a flaw in a Coldcard hardware wallet, involving more than 5200 addresses. It is widely believed that the vulnerability was discovered through malicious use of AI. While the exact details of the discovery mechanism remain controversial in public reports, the incident highlights how quickly attackers can exploit vulnerabilities once they emerge.

In addition, Coinkite CEO Rodolfo Novak warned developers at the time that AI-assisted code review can now spot errors even faster than experienced experts. Variola echoed this reality, saying AI could allow attackers to discover and exploit vulnerabilities at speeds that smaller teams might not be financially matched.

He believes that in practice, protocols may no longer be able to operate without "huge cybersecurity budgets," a restriction that could systematically put small builders at a disadvantage relative to well-capitalized organizations.

Self-hosting, DeFi participation and user burden

瓦里奥拉将影响范围扩展到开发人员团队之外，警告称，一个更具 AI 驱动特征的风险环境可能会让零售用户觉得自我托管和 DeFi 更加危险。他的论点是，攻击越来越将技术突破与针对人类的方法（如社会工程学）相结合。

他说，随着 AI 变得更加普及，用户可能会面临更多设备被入侵或被操纵执行不安全行动的情况。“这使得 DeFi 对零售用户来说吸引力大大降低，因为你必须担心以前不需要那么担心的很多事情。”

这一点尤为重要，因为 DeFi 的增长不仅依赖于流动性和激励措施，还依赖于消费者信心。如果感知的威胁面扩张速度快于安全工具和用户教育的需求，需求可能会转向中心化中介，或者用户完全选择退出链上风险。

防御潜力存在——关键问题是谁能部署它

瓦里奥拉的立场并没有忽视反面论点。此前有安全专业人士警告，AI 既使攻击更加复杂，也有助于加强防御。

今年 4 月，CertiK 高级区块链调查员 Natalie Newson 告诉媒体，“AI 也可以成为最大的防御手段之一”，尽管她谨慎地指出，同样的能力也会提高攻击者的门槛。

This comparison raises a core question in the industry: If AI improves both offensive and defensive capabilities, who has the ability to effectively deploy the defensive side? Variola's comments suggest that there is a gap between the speed of attackers and the budgets of defenders-especially for smaller protocol teams.

What Variola sees as practical value: AI agents are used as assistive, not autonomous

While Variola is skeptical of the overall impact of AI on cryptocurrencies, he acknowledges that there are potential tangible benefits in specific workflows. He pointed out that AI agents can help investors build portfolios or improve trading decisions.

At the same time, he emphasized limitations. Variola said he does not expect AI agents to replace human judgment. "In the final analysis, the final decision is still in the hands of users. So I don't think agents will ever replace the step of taking a trade."

This distinction may be crucial for market participants. AI that provides decision support can reduce user friction, but autonomy introduces new failure modes-especially when models are wrong, miscalibrated, or subject to confrontational manipulation. A support-based approach may reduce these risks while still capturing the dividends of improved efficiency.

展望未来，行业可能需要密切关注两件事：网络安全成本是否继续不成比例地增加，特别是对于较小的协议团队；以及“防御性 AI”的采用是否能在实质上抵消由 AI 驱动的加速攻击。如果平衡继续向大型玩家倾斜，即使 AI 工具在交易所和安全平台中变得越来越普遍，去中心化也可能面临越来越多的压力。