Binance launched Agent OS on Thursday, giving AI agents a way to analyze markets and execute trades on users' behalf. Jeff Li, Binance's vice president of product, said the system keeps control with the user while linking agents to Binance's trading infrastructure.
Li said, "Instead of total freedom, we put the power in users’ hands to give them the granular access control of what they can do through the agent" and added, "We put [the control] at the account level to protect the users’ funds." Binance said it has more than 300 million registered users.
Agent OS and Binance APIs
The launch ties AI applications and agents to Binance APIs, Binance Wallet Agentic Hub, Binance x402 transaction verification and payment facilitator API, and Binance Skill Hub. Binance also added support for Model Context Protocol, and said the system works with OpenAI's ChatGPT and Codex, Anthropic's Claude Code, and Cursor. Through Binance, agents can monitor markets, conduct research and risk analysis, react to signals, and place orders or run strategies such as arbitrage.
That setup gives users a choice in how much autonomy they want to allow. Binance said users can authorize agents to access market data, view account information, and execute trades, then decide whether an agent must ask for approval on every order or can trade on its own once permissions are set.
Sub-accounts and trading limits
Binance said it uses dedicated sub-accounts for agents and that users can assign those accounts to specific activity, including spot trading or futures trading. Withdrawals from those sub-accounts are blocked by default, and Binance does not set a separate cap on how much an agent can trade or lose. The amount a user moves into the sub-account effectively serves as the limit.
That leaves the user to set the boundary before the agent starts trading. For someone testing the system, the practical control point is the transfer into the sub-account and the permission level attached to it, not a firm exchange-wide ceiling.
Binance risk controls
Binance said its existing security, risk-control, and anti-money-laundering policies for subaccount APIs apply to Agent OS at launch. The company said agents can monitor markets and act quickly, but Li said the reasoning behind an agent's decision happens outside Binance's systems.
"We really cannot see the reasoning of what the user’s action is," Li said when asked how Binance can see what leads an agent to make a particular trade. Binance said it can monitor the resulting trading activity, but said it has limited visibility into whether a decision was shaped by faulty information or manipulation.
For users, that means the exchange is overseeing account-level permissions and trade activity, while the agent's internal logic remains outside Binance's view. The unresolved issue is how well those sub-account controls will hold up once agents start placing real trades at scale through Agent OS.
