Major crypto exchanges including Coinbase and Binance are racing to access Anthropic’s Mythos model to defend against AI-powered attacks

Coinbase CSO Philip Martin confirmed the exchange has been in “close communication” with Anthropic about Mythos, while Binance is already testing its systems using commercially available Anthropic models alongside internally developed tools, and Fireblocks says Claude Opus 4.6 has already surfaced vulnerabilities that no prior security audit had found. Crypto firms have largely been excluded from Anthropic's initial Mythos rollout, with Uniswap Labs CEO Hayden Adams publicly asking on X for an Anthropic contact, though some executives expect crypto to receive access before the model is released more broadly. Crypto platforms have good reason to be concerned, as Drift lost $285M this month to hackers who spent months building in-person relationships with its team, Kraken disclosed a criminal group is attempting to extort it after gaining unauthorized access affecting 2,000 accounts, and researchers warn that advanced AI models could compress the timeline for exploiting crypto vulnerabilities from months to minutes.