A new class of AI model from Anthropic is shifting the crypto security focus from isolated code flaws to systemic risks in the underlying infrastructure of decentralized finance.
Back
Anthropic’s Mythos AI forces DeFi to rethink security beyond smart contracts
A New Frontier for Exploits
For years, DeFi security centered on auditing the smart contracts that power protocols like Aave, an open-source liquidity protocol, and Uniswap, a decentralized exchange. While this has matured, Anthropic's Mythos AI is exposing a more dangerous and less understood attack surface: the infrastructure that connects these protocols.
Mythos operates not by scanning for known bugs, but by simulating an intelligent adversary. It explores how small, seemingly unrelated vulnerabilities across different systems can be combined into a multi-step exploit chain. This approach has uncovered weaknesses in the less visible parts of the DeFi stack, including key management systems, oracle networks that feed real-world data to smart contracts, and the cross-chain bridges that allow assets to move between blockchains.
“Composability is what makes DeFi capital efficient and innovative,” Vijender said. “But it also means a minor vulnerability in one protocol can become a critical exploit vector with contagion potential across the ecosystem.”
The Double-Edged Sword of AI
Industry leaders acknowledge the threat but also see AI as a powerful defensive tool. Stani Kulechov, founder of Aave Labs, noted that while AI represents an "evolution in the tools used to achieve exploits," it also equips defenders. Aave has already integrated AI for simulations and code review to complement human auditors.
The key change is the speed and scale of attacks. Manual, pre-deployment audits are no longer sufficient. The new paradigm requires continuous, real-time security analysis, a task well-suited for defensive AI.
“To defend against offensive AI, we will need to take an AI-centric approach where speed and continuous adaptation are essential,” Vijender stated, advocating for systems built with the assumption that breaches will occur.
A Widening Security Gap
The long-term effect may be a divergence in the market. Hayden Adams, founder of Uniswap Labs, expects the gap between secure and insecure protocols to grow. “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.”
Anthropic has recognized the model's power, opting for a controlled rollout under "Project Glasswing" with partners like JPMorgan Chase and Google, rather than a public release. This move aims to give defenders a head start in patching the very vulnerabilities Mythos is designed to find. For the DeFi sector, it marks a fundamental shift from merely securing code to defending an entire, interconnected financial ecosystem against an adversary that thinks at the speed of AI.
This article is for informational purposes only and does not constitute investment advice.
