Monad’s Keone Hon urges crypto teams to adopt AI-driven security tools

Hon’s appeal arrives alongside a comprehensive operational-security guide, which highlights that traditional control-hijacking—such as leaked keys and social engineering—remains a far greater threat than code vulnerabilities. Data from ChainLight’s Lumos tracker supports this, recording $3.31 billion in losses from control-hijacking incidents between January 2023 and the reporting cutoff, dwarfing the $983 million attributed to smart-contract bugs.

The foundation’s guidance emphasizes basic but frequently neglected hygiene: separating personal and work devices, utilizing hardware security keys, and moving away from relying on externally owned accounts for administrative roles. For developers, the Ultrafuzz tool offers a more technical defense. Developed by Antonio Viggiano, the software uses AI agents to identify protocol properties and test whether specific sequences can break them. Monad advises that the tool should be run only on isolated virtual machines, given its need for unrestricted permissions during the investigation process. This push for stricter security follows a string of high-profile breaches, including the compromise of a developer package at Injective and a recent Cloudflare API key leak at Foundation, underscoring the urgency of the foundation's warning.