Ledger Announces AI Security Roadmap

Ledger, a hardware wallet and security infrastructure company, has announced the launch of an AI security roadmap designed for scenarios where AI agents directly conduct financial transactions in the "agentic economy." The company emphasizes that future systems must ensure "AI only proposes, humans only sign," meaning that transaction intents are generated by AI agents, but all key signatures and authorizations must be ultimately confirmed by users via hardware wallets or controlled devices. This structure is explicitly viewed as a core mechanism to prevent "AI Trojan behavior" from automatically executing transactions on high-privilege terminals.

In several public technical blogs, Ledger explains that as AI agents gain broader permissions in browsers, native systems, and financial applications, traditional software security models can no longer withstand automated attacks. The new solution enforces a decoupling of account control and algorithmic decision-making through hardware security modules and cryptographic signing processes, ensuring that AI behavior remains at the "proposal" stage. This roadmap is seen as the first clear endpoint control paradigm in the transition from the "chat AI" era to the "action agent" era.

Source: Public Information