AI trading can be a legitimate use of software, but an “AI-powered” label tells you little about whether a provider is authorised, its performance is real or its service is suitable for you. Treat those as separate questions before connecting an account or sending money.
The US Commodity Futures Trading Commission has warned that fraudsters use AI claims to market trading bots and signal services with unrealistic returns. Its customer advisory explains that AI cannot predict future events or sudden market changes.
The useful response is a verification process. The seven checks below help you examine the business behind the tool, the evidence behind its claims and the access it wants to your account.
First identify what the service actually does
“AI trading” can describe a research assistant, a subscription that sends signals, software that places orders or a service managing money for clients. Those activities create different practical and regulatory questions.
Ask the provider to explain the complete journey from your payment to a trading decision. Who holds the money? Who sends the order? Which company is responsible if something goes wrong?
If you cannot draw that process clearly, pause before considering performance claims.
Seven checks for an AI trading service
1. Verify the legal entity and its permissions
Obtain the provider’s legal name and identify the entity delivering each part of the service. Search the relevant regulator’s own register independently, rather than following a link sent by a salesperson.
For US services, FINRA’s BrokerCheck and the SEC’s Investment Adviser Public Disclosure database are useful starting points for their respective categories. Commodity and derivatives activities can require checks through the NFA’s BASIC database. The appropriate register depends on the activity and jurisdiction; not every software vendor requires the same registration.
Compare the register’s contact details, website and permissions with the business you are dealing with. A genuine firm’s identity can be copied, so a matching name alone is insufficient. Registration also does not guarantee investment performance.
2. Ask where your money will be held
Distinguish paying for software from transferring investment capital to a provider. Ask who legally holds client assets, how withdrawals work and what documentation describes the arrangement.
If a provider claims to work with a recognised broker, verify that relationship directly through the broker’s independently obtained contact information.
An explanation that relies on screenshots of a balance is not a substitute for understanding the account structure.
3. Review account access before connecting a bot
List exactly what the software can do: read balances, view positions, place orders, change settings or withdraw assets.
Use the minimum permissions necessary for the approved function, where the broker supports them. Check how access can be revoked and what happens to open orders when the connection is removed.
Do not assume that disabling withdrawals removes trading risk. Software with trading access can still create substantial losses through orders, leverage or repeated errors.
4. Separate live results from backtests
Ask whether a performance chart reflects simulated trading, a demo account or actual execution with real capital. Request the measurement period, fees, losses and method used to verify the results.
For backtests, ask whether the strategy was assessed on data not used to develop it and whether realistic trading costs were included. A strong historical fit may fail when market conditions change.
A hypothetical example shows why win rate alone is weak evidence. Nine winning trades of $10 produce $90, while one loss of $150 leaves a $60 loss before costs, despite a 90% win rate.
5. Calculate the full cost of using the service
Include subscriptions, commissions, spreads, financing charges and any other applicable fees. Compare results after costs, using the same period and assumptions.
For illustration, a $100 monthly subscription costs $1,200 a year. Against a $5,000 account, that is 24% of starting capital before transaction costs or losses. This calculation does not predict returns; it shows the size of the fee burden.
Ask whether the seller receives referral payments from a broker or exchange. That incentive may affect how the service is promoted.
6. Examine the stop controls
Ask what prevents oversized orders, repeated orders and trading on stale data. Find out who receives incident alerts and whether someone can disable the system promptly.
A simulation can help reveal operational problems without exposing capital, but it does not prove live profitability or execution quality.
| Question | Evidence to request |
| Can order size be limited? | Documented settings and a demonstration |
| What happens if market data stops? | Defined failure behaviour |
| Can the user halt automation? | A tested stop and access-revocation process |
| What happens after an update? | Change notes and reassessment procedure |
These checks examine how a system behaves. They are not assurances that a strategy is safe.
7. Check the regulatory claims separately
“Complies with AI regulations” is not the same statement as “has the permissions required to provide this financial service.” Ask the provider which claim it is making and what supports it.
The AI Global Regulations Atlas can help readers research the broader AI rules in a market. It is not a broker register, an investment-service authorisation check or a certification of a trading bot.
Businessabc.net and Dinis Guarda introduced the resource on November 5, 2026. Their Atlas launch announcement outlines its regulatory coverage. Use that context alongside the relevant financial regulator’s records, rather than treating a country’s AI policy as permission for a provider to handle money.
Warning signs that justify stopping the process
Promises of guaranteed profits, pressure to deposit immediately and refusal to explain losses are reasons to step back. FINRA has also warned about unregistered auto-trading services that exaggerate their AI capabilities and request sensitive brokerage credentials.
If the entity cannot be verified or the account structure remains unclear, a smaller deposit does not solve the problem. Resolve those questions before exposing money or account access.
If you have already shared credentials with a suspicious service, contact your broker using independently verified details, revoke access where possible and preserve messages and transaction records. Report suspected misconduct to the appropriate regulator.
Is AI trading legit or a scam?
AI is a technology category, not a trust rating. Assess the provider, its permissions, the custody arrangement and its evidence separately. Even an authentic service with documented controls can lose money.
The next useful step is to choose one provider claim and verify it independently. If basic facts cannot be confirmed, there is no reason to move on to its return projections.
