The Impact of Anthropic's Mythos Model on Crypto Industry Security

The introduction of Mythos, Anthropic's latest AI model, has sparked significant concern and confusion within traditional tech and finance, and is driving a substantial shift in how the crypto industry approaches security. For years, decentralized finance has focused on defending smart contracts through code audits, vulnerability cataloging, and understanding common exploits. However, Mythos, designed to identify and chain weaknesses across systems, is pushing attention beyond code and into the underlying infrastructure. According to Paul Vijender, head of security at Gauntlet, a risk management firm, 'The bigger risks lie in infrastructure.' He emphasized that when considering AI-driven threats, his primary concern is not smart contract exploits but rather AI-assisted attacks against human and infrastructure layers. This includes key management systems, signing services, bridges, oracle networks, and cryptographic layers that connect them. These components are less visible than smart contracts and often fall outside traditional audit scope. Recently, web infrastructure provider Vercel disclosed a security breach that may have exposed customer API keys, prompting crypto projects to review their code and rotate credentials. The intrusion was traced to a compromised Google Workspace connection via the third-party AI tool Context.ai. Mythos represents a new class of AI systems built to simulate adversaries, exploring how protocols interact and testing how small weaknesses can be combined into real-world exploits. This approach has drawn attention beyond crypto, with banks like JP Morgan treating AI-driven cyber risk as systemic and exploring tools like Mythos for stress testing. Early findings from models like Mythos have identified weaknesses in behind-the-scenes systems that keep crypto platforms secure, including technology that protects keys and handles communication between systems. Paul Vijender noted, 'I think there are two areas where AI models are especially valuable: multi-step exploit chains that historically only get discovered after money is lost, and infrastructure-layer vulnerabilities that traditional audits never touch.' This shift matters in a system built on composability, where DeFi protocols connect and build on each other's services. DeFi protocols are designed to interconnect, sharing liquidity and relying on common oracles, which creates pathways for risk to spread. Without AI, these dependencies are hard to trace; with AI, they can be mapped and exploited at scale, resulting in a shift from isolated exploits to systemic failures that cascade across protocols. The evolution of AI attacks has led some industry leaders to view Mythos as an acceleration rather than a turning point. Stani Kulechov, founder of Aave Labs, stated that AI reflects the dynamics already at play in DeFi's adversarial environment. Kulechov believes that 'Web3 is no stranger to well-funded and motivated adversaries,' and AI models represent an evolution in the tools used to achieve exploits. DeFi is already built for machine-speed attacks, with smart contracts executing automatically and defenses operating without human intervention. Aave is seeing AI surface new categories of vulnerabilities, including issues that human auditors may have previously deprioritized. The Mythos paper shows that AI can uncover old bugs that were previously deprioritized. If attackers can move faster, the question becomes whether defenses can keep pace. For both Gauntlet and Aave, the answer lies in changing the security model itself. Audits before deployment and monitoring after were designed for human-paced threats; AI compresses that timeline. To defend against offensive AI, a shift towards an AI-centric approach where speed and continuous adaptation are essential is necessary. This includes continuous auditing, real-time simulation, and systems built with the assumption that breaches will happen. Aave has integrated AI into its workflows, using it for simulations and code review alongside human auditors. 'We take an AI-first approach where it adds clear value,' Kulechov said, 'but it complements, rather than replaces, human-led auditing.' AI equips both attackers and defenders, and for builders, the long-term effect may be less disruption than divergence. Hayden Adams, founder and CEO of Uniswap Labs, stated, 'AI gives builders better ways to stress test and harden systems.' Over time, Adams expects the gap between secure and insecure protocols to widen. 'Projects that prioritize security will have greater ability to test and harden systems before launching,' he said. 'Projects that don’t will be most at risk.' This may be the real shift: security is no longer about eliminating vulnerabilities but about continuously adapting to a system in which those vulnerabilities are constantly rediscovered and recombined.