Is Safe Smart Account safe to connect to an AI agent?

What an AI agent can do

• Live vs paper: live; contracts are deployed on many networks (see Safe's supported-networks list) source
• Order approval: open-source multisig smart account: Safe's AI-agent guide makes the agent one signer with a threshold that needs at least one human approval; an allowance module can give the agent a per-token spending limit that resets on a schedule; guards can check every transaction source
• Markets: EVM chains
• Pricing: open-source contracts (LGPL-3.0); network gas fees apply

Safeguards that exist

• Paper, demo or testnet mode: The agent quickstart runs on the Sepolia testnet (chain ID 11155111) (official docs) source
• Withdrawal permission needed or allowed: Documented setup makes the agent one signer in a 2-of-3 Safe, so it can propose but cannot execute transfers without human approval (official docs) source
• Scoped keys, permission scopes or IP whitelisting: Safe supports adding spending limits, timelocks and whitelists via modules to restrict agents (official docs) source
• Open source or audited: Open source source

Safeguards that are missing

• Can it place live orders unattended?: Can place live orders with no documented approval step source
• Order approval or confirmation options: No approval or limit option in our listing source

Sources

• https://docs.safe.global/home/ai-agent-quickstarts/human-approval
• https://docs.safe.global/home/ai-overview

Full listing: Safe Smart Account · How the score works: Safety score method