Bybit, the world’s second-largest cryptocurrency exchange by trading volume, has rolled out AI Sub-Accounts, a dedicated account type designed to improve security and user control as AI agents become increasingly integrated into crypto trading workflows.
The feature introduces a separate environment specifically built for AI-powered trading activity. Unlike regular sub-accounts, custodial accounts, or Islamic accounts, AI Sub-Accounts are structured to limit the operational scope of automated agents while keeping primary user funds protected from potential AI-related risks.
The launch reflects growing concerns around unrestricted API access and the rapid expansion of agentic trading systems across digital asset markets.
Addressing AI agent risks
As AI trading agents become more common, exchanges and traders face new security challenges tied to automation. Compromised AI models, flawed trading logic, or malicious third-party agents can potentially trigger unauthorized trading activity, excessive leverage exposure, or unintended liquidation events.
Bybit said the new account architecture creates a clear separation between human account control and machine execution.
"We recognize that as agentic trading enters the mainstream, the security baseline has to evolve. No agent should have unchecked power over a trader’s full portfolio," said Victor Wu, Head of AI Agent Architecture at Bybit.
According to the company, all AI-driven activity inside the system remains fully ringfenced from users’ main accounts and other sub-accounts.
The AI Sub-Account framework introduces multiple restrictions designed to contain automated activity within tightly controlled boundaries.
AI agents operate inside isolated accounts that cannot directly access primary funds or move assets across other accounts. Users can configure limits on leverage, asset exposure, transfers, and authorized commands before connecting trading agents.
The system also includes read-only oversight features, allowing traders to monitor all AI-driven activity in real time without giving agents broader account permissions.
Bybit added that the structure is intended to help users safely test new AI agents or experimental trading strategies before deploying them more broadly.
AI infrastructure becoming core exchange feature
The launch highlights how exchanges are increasingly redesigning infrastructure around AI-native trading behavior rather than treating automation as an optional add-on.
As AI systems begin moving beyond market analysis into strategy execution and portfolio management, exchanges are facing growing pressure to build architecture capable of balancing automation with risk containment.
Bybit said all users connecting AI agents to the platform will now operate through AI Sub-Accounts by default, establishing a baseline protection layer regardless of user experience or technical expertise.
The move positions AI security as a core component of future exchange infrastructure as competition around automated trading ecosystems continues to intensify.
Dan Burgin
