Interactive Brokers lets AI trade without per-order approval

On 21 October Interactive Brokers announced IBKR Autonomous Trading. Eligible clients can authorise connected AI platforms to monitor markets and place orders on their behalf within rules they set, across more than 75,000 stocks, options and ETFs worldwide.

IBKR launched AI integrations with Claude in June, then ChatGPT and Grok. Until now every AI-generated order waited in a tab for client approval under the principle "the client approves every trade". The new feature lets eligible clients remove that step, turning AI from adviser into authorised executor.

Orders go through a dedicated AI segment funded by transfers from the main account, so the AI only reaches the amount set aside. No new account is needed. Clients define strategies and limits in natural language; the AI can monitor criteria, manage portfolio risk and place stock and stock-option orders when conditions are met.

Eligibility depends on entity, country, regulation and account type, with funding and account limits at launch. The release did not name which AI platforms support the autonomous mode. CEO Milan Galik: "Clients define their trading strategies and safeguards; the connected AI monitors markets and executes within those parameters."

Giving agents authority over money is the industry's most contested threshold. It is also a behavioural finance experiment: will agents avoid emotional errors, or tirelessly execute a badly defined rule?

The biggest risk is an AI executing a flawed or ambiguous instruction quickly and repeatedly. A person might catch an error after a few trades; an agent can run the same rule dozens of times in seconds. Prompt injection is another threat: instructions hidden in a news item or social media post the agent reads could steer it into the wrong trade. The dedicated balance and launch limits are designed to contain these risks.

• Start with a small allocation.
• Write rules in numbers: entry, exit, stop-loss.
• Review every trade and rationale weekly.
• Be transparent with clients about which decisions AI makes alone.