CoinDesk
Anthropic's Mythos model is reshaping how the DeFi industry thinks about security, pulling the focus away from smart contract code and toward the infrastructure underneath it — key management systems, bridges, oracle networks and the signing services that connect them. Early findings from the model have surfaced weaknesses in those behind-the-scenes layers that traditional audits never touch, and both Coinbase and Binance have reportedly approached Anthropic to test it. Banks including JPMorgan are treating AI-driven cyber risk as systemic enough to warrant stress-testing tools like Mythos.
Why it matters
DeFi's defenses were built around auditable code; Mythos is built to simulate adversaries. Instead of scanning for known bugs, it explores how protocols interact, chaining small weaknesses into real-world exploits. In a system wired together by composability — shared liquidity, common oracles, layered integrations — a minor flaw in one protocol can become a contagion vector across the ecosystem. The Hyperbridge exploit, in which an attacker minted $1 billion worth of bridged Polkadot tokens on Ethereum by exploiting a flaw in cross-chain message verification, is the kind of cascading failure the new model is designed to surface before money moves.
Market impact
Protocols are already restructuring around the threat. Aave Labs has folded AI into its simulations and code review alongside human auditors, and Uniswap's Hayden Adams told CoinDesk the lab is watching Mythos closely. Gauntlet's Paul Vijender framed the shift bluntly: defending against offensive AI requires continuous auditing, real-time simulation and systems built on the assumption that breaches will happen. The longer-term read from the people building on top of the rails is divergence, not disruption — projects that prioritize security will harden faster, and the gap between them and the rest will widen.
Frequently asked questions
-
What is Anthropic's Mythos model and why does it matter for DeFi?
Mythos is an Anthropic AI model designed to simulate adversaries by mapping how protocols interact and chaining small weaknesses into real-world exploits. It matters for DeFi because the system is built on composability, where a minor flaw in one protocol can cascade across the ecosystem.
-
Why are DeFi leaders worried about infrastructure risks more than smart contract bugs?
Smart contracts have been audited for years and common exploits are well cataloged. Infrastructure layers — key management, bridges, oracle networks, signing services — are less visible and often outside traditional audit scope, making them attractive targets for AI-driven attacks.
-
What recent exploits illustrate the cascading-failure risk Mythos aims to surface?
The Hyperbridge attack, in which an attacker minted $1 billion worth of bridged Polkadot tokens on Ethereum by exploiting a flaw in cross-chain message verification, is cited as the kind of systemic, multi-step failure the new model is designed to flag before money moves.
-
How are major exchanges and banks responding to Mythos?
Coinbase and Binance have reportedly approached Anthropic to test Mythos, and JPMorgan is treating AI-driven cyber risk as systemic enough to stress-test against using tools like the model.
-
How are DeFi protocols adapting their security models to AI-driven threats?
Aave Labs has integrated AI into simulations and code review alongside human auditors, while Gauntlet advocates continuous auditing, real-time simulation and systems designed on the assumption that breaches will happen. The longer-term expectation is divergence — projects that prioritize security will harden faster…
TheBlock
CoinTelegraph
WatcherGuru
WuBlockchain