AI trading tools are real and legal. Much of what is sold as AI trading is not. How to tell the two apart, what US regulators warn about, and what the research says about bots making money.
Updated November 16, 2026
14
min to read
The short answer
AI trading software is real and legal: you can use it to read charts, test rules on past data and place trades in an account that stays in your name. But many products sold as AI trading are scams, and no tool can promise profits. Promised returns, pressure to deposit and trouble withdrawing are the classic warning signs.
Yes on both counts. Software that reads charts, tests trading rules on past prices, sends alerts or places orders for you is real, and using it for your own trading is legal in the US. What the law regulates is who touches your money and who advises you. Firms that manage money, give investment advice or run a trading platform generally have to be registered, which is why the SEC, FINRA and the state regulators in NASAA tell investors, in a joint alert on AI fraud, to treat a missing registration as a prompt to dig further before investing anything.
The trouble is the label. “AI” sells, so it gets stuck on everything from a simple rule-based bot to an outright fraud. FINRA calls false or inflated AI claims AI washing. The SEC has charged firms for it: in October 2024, Rimar Capital and its owner settled charges for describing “an AI-driven platform for trading securities” it did not have, after raising nearly $4 million from 45 investors. The SEC’s AI washing cases against investment advisers are covered in our guide to AI investing.
So the useful question is not “is AI trading legit?” but “is this product legit?” Most fall into one of three groups. For what an AI trading agent is and how it differs from a bot, see our guide to AI trading agents.
| What it is | Who holds your money | Who decides the trades | What to check |
|---|---|---|---|
| A tool you use: chart reads, backtests, alerts, orders you set up | You, in an account or wallet in your name | You, or rules you wrote and switched on | What it can do with your account, and whether it shows its work |
| A managed or auto-trading service | Your brokerage account, traded by someone else | The provider | Its registration, on Investor.gov or FINRA’s BrokerCheck |
| A “platform” you send money to, where AI supposedly trades | Them | Often nobody: in the fraud cases below, little or no trading happened | Nothing to check. This is the pattern regulators describe in AI trading fraud |
Regulators have described the same playbook for years. The costume changes, from forex robots to crypto arbitrage to AI, but the steps don’t.
A text, a social media ad, a dating app or an “investment club” in a group chat. The FBI says crypto investment scams usually start this way and move the conversation to a messaging app.
A bot or algorithm that supposedly wins almost every trade. The CFTC has seen claims of returns of “tens of thousands of percent” and 100% “win” rates.
Not to a broker or exchange you chose and can look up, but to a website or app they control.
The numbers are fake. The FBI describes victims being shown fake profits and offered loans to invest more.
Suddenly there are “taxes” or “fees” to pay first. Paying them doesn’t free the money; it adds to the loss.
Real cases follow that script:
AI also makes the old tricks more convincing. In its 2025 report, the FBI counts more than $632 million lost in investment complaints with a reported AI link, from chat scripts written by AI to fake videos of celebrities and CEOs, and notes that many victims don’t realize how much AI was involved. The FTC, which collects its own reports, puts losses to investment scams in 2025 at more than $7.9 billion, with a median loss above $10,000.
Any one of these is a reason to stop. Two or more is the pattern regulators keep describing.
| Red flag | What it sounds like | Who warns about it |
|---|---|---|
| Promised or steady returns | “10% a month”, “a 100% win rate”, “AI picks the winners” | CFTC, FINRA, SEC |
| Your money goes to them | “Deposit on our platform”, “send USDT to this wallet” | FBI, SEC |
| Pressure | “Only today”, “the spots are almost gone”, a loan to invest more | SEC, FBI |
| No registration you can check | A vague or offshore company, “licensed” with no name or number | SEC, FINRA, NASAA |
| Trouble withdrawing | A “tax” or “fee” before you can take money out | FBI, SEC |
| Proof you can’t verify | Screenshots of gains, testimonials, a celebrity video | FINRA, SEC |
| Vague AI | “Proprietary AI” with no explanation of what it does | FINRA |
| It wants your login | Your brokerage username and password | FINRA |
Checking takes a few minutes. Look up the firm and the person on Investor.gov and FINRA’s BrokerCheck, the free tools the SEC and FINRA point to. If an app says it works with a known broker, contact that broker using details you found yourself, as FINRA suggests. The CFTC suggests checking how old the website is with a domain lookup at lookup.icann.org, and a reverse image search on the people in the pitch. A registration tells you who is accountable, not that a strategy works.
Some do, some of the time. Here is what the research says about the rest, and why AI doesn’t change it.
A standard test of trading rules is to run them on long price histories. Pierre Bajgrowicz and Olivier Scaillet tested technical trading rules on the Dow Jones Industrial Average from 1897 to 2011. They found that an investor could never have picked the future best rules in advance, and that even in the past, the rules’ performance “is completely offset by the introduction of low transaction costs.”
A backtest shows what a rule would have done on past prices. Researchers at Quantopian, a platform where users wrote and tested trading algorithms, compared 888 algorithms with their later results. The backtest Sharpe ratio, a standard score of return per unit of risk, explained almost nothing about what came next, and the more a strategy had been backtested, the bigger the gap between its backtest and what followed. That is overfitting: a rule tuned to the noise of the past. Our guide to backtesting a trading strategy shows how to avoid it, with a real example.
Bots make it easy to trade a lot. In a classic study of 66,465 US households, Brad Barber and Terrance Odean found that those who traded the most earned 11.4% a year from 1991 to 1996, while the market returned 17.9%. Every trade pays a spread and a fee, and trading often multiplies them.
A model that reads more data still pays the same costs, faces the same competition and meets markets that change. The CFTC puts it plainly: “AI technology can’t predict the future or sudden market changes.” So the honest answer to “can an AI bot make money?” is that a well-built one can make money in some markets and periods and lose it in others. Anyone who says it will make money every month is the red flag. What the research says about AI forecasts is in our guide to whether AI can predict the stock market.
Put any product through these six questions. A legit tool answers them plainly, in writing.
| Ask this | A legit tool | A scam |
|---|---|---|
| Where is my money? | In an account or wallet in your name, at a venue you can look up | On “their platform”, or sent to a person |
| What can it do with my account? | It says so: read data, place trades, move money, each one stated | Vague, or it asks for your password or private key |
| Does it show its work? | The rules, the backtests, every trade, the losing stretches | Screenshots of wins and a secret “AI” |
| What does it promise? | Nothing about returns, and it states the risks | A percentage per month, or a win rate near 100% |
| Can I leave? | Withdraw and cancel when you like | “Fees” or “taxes” to get your money back |
| Who is behind it? | A named company you can look up, registered where the law requires it | Unnamed, unregistered or impossible to verify |
Before you connect any tool to an account with money in it, get written answers to four more: what permissions does it ask for, what can it do without asking you, is every action logged where you can see it, and how do you switch it off? FINRA adds two habits worth keeping. Any automated trading service “should be able to explain with specificity how its technology works,” and even with a registered one, you should keep watching the activity in your account.
A legit bot removes the fraud risk, not the market risk. Three things can still go wrong:
The careful way to start is small and slow: test the rule on past data, run it on paper or with a small amount, set your limits first, and read every trade it makes. If the tool trades on Hyperliquid, is Hyperliquid safe? covers the venue’s own risks. If you are wondering whether a chatbot can do this for you, here is what Claude can and can’t do for a trader, and whether ChatGPT can trade for you. For what a legit AI trading app looks like, see Ethy’s own page.
People add “reddit” to the search because they want someone with nothing to sell. That instinct is right, but forums and videos have their own distortions:
What does count: a long, complete record with the losing trades in it, a method you can understand, and a test you can repeat yourself.
We build Ethy, an AI trading agent, so read this part knowing that. Here is what it does, so you can run it through the checklist above yourself.
Yes. Using software, AI included, to analyze markets or place your own trades is legal. Firms that manage your money, give investment advice or run a trading platform generally have to be registered with the right regulator. The SEC, FINRA and NASAA say a missing registration is a reason to investigate further before you invest any money.
Not reliably. The CFTC’s advisory says it plainly: AI technology can’t predict the future or sudden market changes. AI can help you process information faster, test rules on past data and watch markets for you, but it doesn’t know what happens next. A product that claims it does is making the exact claim regulators list as a red flag.
A fake platform. Someone you met online, often in a group chat, says an AI system trades for members. You deposit on their site or app, the dashboard shows gains, and withdrawals are blocked by “fees” or “taxes”. The FBI, the SEC and the CFTC describe this pattern in recent cases. Little or no trading happens; the money is simply taken.
Search the firm and the people on Investor.gov and FINRA’s BrokerCheck, and for futures or commodity products use the CFTC’s registration check. If an app claims a partnership with a known broker, contact the broker through details you looked up yourself. Check the website’s age at lookup.icann.org. Registration makes someone accountable; it doesn’t make a strategy profitable.
Stop sending money, including any “fee” to unlock a withdrawal. Report it to the FBI’s Internet Crime Complaint Center, and to the SEC, the CFTC or FINRA depending on the product. Keep every message and transaction record. Be wary of anyone who offers to recover your money: the FBI warns that victims are targeted again by recovery scams.
Judge it by the checklist above. Ethy is made by ETHY AI LTD, a company incorporated in England and Wales (number 16530519). It is software for research and analysis, doesn’t give investment advice and doesn’t promise returns. Your balance lives in a wallet that belongs to you, and you can export its key at any time. You can try it on the free plan first.
Trading involves risk, including the loss of your money. Ethy AI is a research and trading tool and does not give investment advice or recommendations. Backtests are simulations on historical data and do not predict future results.
Deploy your AI trading agent in one click.
