Anthropic announced the launch of two next-generation AI models—Claude Mythos 5 (security-focused) and Claude Fable 5 (general public)—on June 9 (local time), triggering heightened security concerns across the DeFi sector. The announcement coincided with a whale wallet holding approximately $300 million worth of KHYPE moving funds and using RevokeCash, prompting warnings from on-chain analyst Emmett Gallic about potential smart contract exploit automation. The concerns stem from Anthropic's April disclosure that a preview version of the model identified thousands of software vulnerabilities, with the company acknowledging risks of attackers developing new hacking tools if the technology is misused.
Whale Wallet Moves $300M KHYPE Following Model Announcement
On-chain analyst Emmett Gallic reported on X that an address holding approximately $300 million worth of KHYPE used RevokeCash after transferring the majority of funds out of the wallet. The timing of the transaction aligned with Anthropic's public release of Claude Mythos 5 and Claude Fable 5, amplifying concerns about smart contract exploit possibilities. KHYPE is a wrapped and staked asset based on HYPE, the native token of decentralized derivatives platform Hyperliquid. RevokeCash is a service that allows wallets to revoke token usage permissions (Approve) previously granted to specific smart contracts.
Anthropic Discloses Security Model Capabilities in April Preview
Anthropic announced on June 9 (local time) the planned release of Claude Mythos 5, a security-specialized AI model, and Claude Fable 5, a general-purpose model for public use. In April, the company disclosed that a preview version of the model detected thousands of software vulnerabilities. At that time, Anthropic expressed concerns that malicious actors could exploit the model to develop new hacking tools if misused.
Chainalysis Reports AI-Enabled Smart Contract Analysis Risks
Blockchain analytics firm Chainalysis stated in a report released on June 9 (local time) that AI advancements have enabled attackers to analyze unverified smart contracts more easily. The firm noted that hacks in 2026 targeting Truebit, Trusted Volumes, Aperture Finance, and Ekubo all exploited unverified smart contracts with undisclosed source code. Chainalysis highlighted that while skilled engineers previously required days to analyze a single smart contract, such tasks can now be partially automated using AI.
FAQ
What AI models did Anthropic announce on June 9?
Anthropic announced Claude Mythos 5, a security-focused AI model, and Claude Fable 5, a general-purpose model for public use, on June 9 (local time).
Why did the whale wallet use RevokeCash after moving $300M in KHYPE?
The wallet used RevokeCash to revoke token usage permissions granted to specific smart contracts, a precautionary measure that coincided with the launch of Anthropic's AI models capable of detecting software vulnerabilities.
What vulnerabilities did Anthropic's model identify in April?
In April, Anthropic disclosed that a preview version of its model detected thousands of software vulnerabilities, raising concerns about potential misuse by attackers to develop new hacking tools.
