5 questions to ask any AI trading platform before depositing
Most AI-trading marketing is designed to be skimmed. The questions below are designed to be answered in writing. If a sales agent — human or otherwise — refuses to put the answers on the record, treat that refusal as the answer.
1. Which regulator licenses you, and what does the licence cover? Demand the legal entity name, the licence number, and the regulator’s register URL. Then verify the licence yourself. Many “regulated” platforms hold a licence that does not cover the AI products they are selling you.
2. What does the AI actually do — recommend, execute, or both? The legal and practical implications of each are very different. A recommendation tool can be wrong without anyone trading; an execution system can be wrong while you sleep. The platform should be precise about which mode it operates in for each feature.
3. How is the model evaluated, and on what data? You are looking for an out-of-sample test period, a clear cost model that includes spread and slippage, and a description of the regimes the model has and has not seen. A platform that cannot describe its evaluation methodology in two paragraphs has not done it properly.
4. What happens when the model is wrong? Every model is wrong sometimes. The right answer here is a defined risk envelope — maximum position size per signal, maximum drawdown before the system pauses, and a kill-switch you can use yourself. Vague language about “risk management” is not an answer.
5. What does my full all-in cost look like for one round-trip? Spread, commission, financing, FX conversion and withdrawal — added together for a representative trade. If the platform refuses to give you a worked example, do the experiment yourself with the smallest deposit they accept. The number that comes out is your real cost.
A platform that answers all five questions cleanly in writing is not necessarily a good platform — but a platform that cannot answer them is reliably a bad one. Use the list before depositing, not after.
Related insights
Why 2026 is the year retail AI trading goes mainstream
A confluence of cheap inference, regulator clarity and platform consolidation is dragging AI trading out of the niche and into ordinary brokerage accounts.
How regulators are responding to AI brokers
A short, jurisdiction-by-jurisdiction tour of how the major regulators are framing AI features in retail trading — and where the rules still have gaps.