Hardware wallet provider Ledger has introduced Ledger Agent Stack, an open-source toolkit designed to let AI agents interact with cryptocurrency wallets and blockchain tools while maintaining rigorous human oversight. The launch aims to strike a balance between automation and security, preventing rogue AI transactions by requiring explicit physical confirmation on a Ledger device for any sensitive action.
Under the new framework, AI bots can read wallet balances, suggest transactions, prepare token swaps, and draft operations, but they cannot execute anything without human approval. This contrasts with some existing solutions that permit agents to act autonomously within preset rules. Ledger’s conservative approach is rooted in data: the company notes that 26.1% of all AI agent skills contain at least one security vulnerability, while human error accounts for roughly 60% of all security breaches. A hybrid model, where AI proposes and humans approve, is intended to mitigate both vectors.
“By anchoring every signing decision in hardware, Ledger Agent Stack is designed to ensure that even if an agent’s software environment is compromised, funds cannot move and sensitive information cannot be accessed without explicit human confirmation on your Ledger signer,” the firm stated.
Beyond crypto, Ledger is positioning its devices as universal security keys. Support for the OpenPGP encryption standard allows developers to protect API keys and AI agent credentials, while the hardware can also emulate YubiKey-type authentication for platforms like GitHub, npm, 1Password, and Discord. MoonPay and Shisa.ai have already integrated the Agent Stack into their products.
The release forms part of Ledger’s 2026 AI Roadmap, with upcoming tools for agent identity, execution, and “proof of human” attestation. Founded in 2014 and headquartered in Paris, Ledger is reportedly exploring an initial public offering that could value the company at more than $4 billion.
