Crypto users wary as Anthropic releases Claude Mythos with safeguards

Crypto Insight Summary: Anthropic released the first public version of its Claude Mythos AI model, called Fable 5, on June 10, 2026, with built-in safeguards that route cybersecurity-related queries to a separate model, Claude Opus 4.8. Last month, Anthropic disclosed that Mythos had identified over 10,000 high or critical-severity vulnerabilities in systemically important software through Project Glasswing, including roughly 6,200 in more than 1,000 open-source projects relevant to crypto. The release sparked concern in the crypto community, with Moonrock Capital founder Simon Dedic warning that the model would reduce the cost and skill required to find smart contract exploits to "basically zero," urging users to revoke wallet approvals and move funds to fresh hardware wallets. Curve Finance co-founder Michael Egorov pushed back, arguing that Mythos's success in auditing large codebases may not translate to DeFi smart contracts, which have far fewer lines of code and are already manageable for human and conventional AI auditors. Egorov suggested the more likely threats were operational security failures, such as multisig key compromises and supply chain attacks on front-end dependencies. The concerns come amid a surge in AI-driven crypto attacks, with April 2026 hack losses reaching $629.7 million, the highest since February 2025. A separate, less restricted version called Claude Mythos 5 has been made available to a small group of cybersecurity and infrastructure providers with some safeguards lifted. Crypto Insight Analysis: Direction: Bearish Crypto Impact: Major Impact Horizon: Short-term (1-3 days) Reasoning: AI model release raises smart contract security concerns, increasing market fear and potential sell pressure.