How Does Crypto Remain Secure in a World of Always On AI Hacks? - Uneasy Money

In this episode of Unchained's Uneasy Money, host Cain Warwick and co-host Taylor Monaghan are joined by Austin Griffith of the Ethereum Foundation to explore the growing threat of AI-powered hacking, particularly in light of Anthropic's newly released 'Mythos' model. The conversation centers on how AI like Mythos—capable of identifying 20 zero-day vulnerabilities in legacy software—could revolutionize both cyberattacks and security, with implications for crypto infrastructure. The hosts express concern that AI is no longer just assisting hackers but acting autonomously, potentially rendering traditional security measures obsolete. They discuss the paradox of crypto’s immutability, which offers security but also creates long-term risks when flaws are discovered years after deployment. The episode also dives into practical implications: the rise of 'skill files' that instantly upgrade AI agents with domain-specific knowledge, the cost trade-offs between frontier models like Mythos and cheaper local alternatives, and the need for smarter model routing systems. A major theme is the shift from human-led development to AI-driven automation, where even DevOps and infrastructure management are now being handled by agents. The hosts also critique Anthropic’s move to restrict access to OpenClaw via API-only models, suggesting it’s driven by profit and control rather than security, and speculate on the potential for AI-powered insider trading and front-running on platforms like Polymarket. Despite the chaos, they remain optimistic that the ecosystem will adapt through better tooling, skill-based AI, and strategic cost optimization.