How to Spot AI-Washing in Crypto: 9 Red Flags

How to Spot AI-Washing in Crypto: 9 Red Flags

Artificial intelligence has become the marketing engine of the current crypto cycle. Some of that excitement is justified: machine learning genuinely powers fraud detection, on-chain analytics, and trading research. But alongside the real work, a predatory pattern has emerged. It is called AI-washing, and it involves dressing up an ordinary Ponzi scheme, a copy-trading bot, or a thin script in the language of "proprietary neural networks" to justify impossible returns.

This is not a fringe concern. Regulators have already started naming and penalising it. In March 2024, the U.S. Securities and Exchange Commission charged two investment advisers, Delphia and Global Predictions, with making false and misleading statements about their use of AI, in what it described as its first "AI-washing" enforcement actions. Months later, the Federal Trade Commission launched a sweep it called Operation AI Comply, targeting companies that "used AI to further deceptive or unfair conduct."

Scepticism is your best protection. When a platform claims its bot never loses because of a secret algorithm, treat that as a warning, not a reassurance. This guide gives you a calm, repeatable framework to tell genuine innovation apart from marketing built to separate you from your money.

A note on uncertainty: the figures below come from named regulators and researchers and are clearly attributed. Scam tactics evolve quickly, and no checklist catches every case. Treat these red flags as cumulative signals, not a guaranteed test.

What is AI-washing?

AI-washing is the practice of overstating, or entirely inventing, the role of artificial intelligence in a product. The term echoes "greenwashing," where companies exaggerate environmental credentials. The FTC frames it as misleading consumers "about whether a service or solution actually uses AI technology," exaggerating "what an AI product can do," or seizing "on AI hype to lure consumers into bogus schemes."

In crypto specifically, AI-washing tends to do one of three things:

• Inflate a token. A project bolts "AI" onto its name and whitepaper to ride a narrative and pump the price, with little or no working technology behind it.
• Justify impossible yield. A platform claims an AI trading bot delivers steady daily profits, masking what is really a Ponzi structure paying old depositors with new deposits.
• Create a black box. Vague references to deep learning discourage you from asking how the money is actually made, which is exactly the question a scam cannot survive.

If you do not understand how a platform generates returns, you cannot assess its risk. That is the whole point of the disguise. The nine sections that follow break the disguise down into observable, checkable signals.

Quick-reference: AI-washing red flags

Use the table as a triage tool. A single flag warrants caution. Two or more, especially any marked "severe," means walk away.

1. Fixed daily returns without drawdowns

Financial markets are volatile by nature. Every real trading strategy has days when the market moves against it. If a platform promises a consistent return every single day, week after week, it is almost certainly not trading at all. It is paying earlier investors with money from newer ones.

A genuine AI model is still subject to market conditions. It has winning streaks and losing stretches. A perfectly smooth, always-up equity curve is a mathematical near-impossibility in live crypto markets. The UK's Financial Conduct Authority warns that fraudsters running fake investment platforms often manipulate software to fake prices and returns so the dashboard looks profitable until you try to withdraw.

If the pitch is "guaranteed" or "risk-free" daily profit, the only thing guaranteed is your exposure to loss. Compare what you are being shown against the honest, mixed performance disclosures on a regulated venue. Our look at the best UK crypto exchanges for 2026 sets a useful baseline for what a transparent platform actually discloses.

2. The absence of audited performance data

Legitimate quantitative firms publish a verifiable track record: entries, exits, fees, and net profit. AI-washed platforms typically show only a rising balance on a slick dashboard, with no underlying data and no way to confirm it.

Without independent verification, those numbers are just pixels. Ask for:

• A trade-by-trade log you can inspect.
• A live, read-only connection to a public performance tracker.
• A reference to a reputable, named auditor, not a logo with no link.

If none of that exists, assume the performance is fabricated. The SEC's action against the two advisers turned precisely on the gap between what was claimed and what could be substantiated, and the firms agreed to pay civil penalties totalling 400,000 dollars without admitting or denying the findings.

3. Vague technical descriptions

Ask a real engineer how their model works and they can describe the inputs and the method: maybe sentiment analysis, maybe mean reversion, maybe a specific class of model on a specific dataset. AI-washed projects retreat into empty phrases such as "advanced neural networks," "quantum AI," or "proprietary deep learning" with no detail behind them.

When technology cannot be explained in plain language, that is usually because it does not exist. Scammers rely on the assumption that AI is too complex for ordinary people to question. Flip that assumption. The inability to give a clear, jargon-free answer is itself the red flag. Watchdog warnings such as our YieldMax AI scam warning show how thin the substance behind the buzzwords often is.

4. Aggressive referral and affiliate schemes

A business that genuinely profits from trading does not need to pay you to recruit your friends. When a platform's core incentive is building a "team" or "downline," the real product is recruitment, and that is the defining structure of a pyramid scheme.

The maths rarely works either. If a large share of each deposit is paid out in referral commissions, the "AI" would need to perform heroically just to break even before anyone profits honestly. High recruitment bonuses are a strong signal that money is flowing between participants rather than being earned in a market. Our CryptoMine Pro scam warning documents how an "AI mining" pitch leaned on exactly this kind of referral engine.

5. Anonymous or unverifiable founders

Innovation requires accountability. People who genuinely build AI systems tend to leave a footprint: published work, conference talks, employment history, named institutions. If a project's founders are anonymous, use generic stock photos, or surface only on profiles created weeks ago, you have no recourse when the project disappears.

Do the basic checks:

• Search the founders' names alongside their claimed employers and credentials.
• Look for a professional history that predates the project.
• Be wary of profiles with no genuine endorsements from real, identifiable people.

Anonymity is not always malicious in crypto culture, but combined with promises of guaranteed returns, it removes the one thing that lets victims pursue justice later. Chainalysis has documented how AI tools now help fraudsters generate realistic fake websites, listings, and personas at scale, which makes a fabricated founder easier to manufacture than ever.

6. Reliance on screenshots over live data

Screenshots are trivially faked. A common tactic is to flood a Telegram channel or X feed with images of huge percentage gains. These are frequently pulled from demo accounts or edited to hide the losing trades.

Demand live, read-only access to the trading account, or a verifiable third-party link that you, not the promoter, control. If a platform refuses, it is hiding something. The FCA's guidance specifically warns that scam adverts lead to "professional-looking websites" where the numbers are simply made up, so do not let production values stand in for evidence.

7. Requests for withdrawal permissions

This is one of the clearest technical tells. A genuine trading bot only needs permission to buy and sell on your behalf. It should never need the ability to withdraw funds from your exchange account.

If a platform asks you to create an API key with withdrawal rights enabled, it can drain your balance instantly. Before connecting anything:

• Open your exchange's API settings and create a trade-only key.
• Confirm the "Enable Withdrawals" permission is switched off.
• Consider IP-whitelisting the key so it only works from a known address.

For longer-term holdings, the safest answer is to remove them from any bot's reach entirely. Moving assets into a hardware wallet isolates them from malicious scripts and connected platforms. Our guide to self-custody versus custodial wallets for 2026 explains the trade-offs and how to set this up. If you are comparing reputable venues for the funds you do keep on an exchange, our Kraken review walks through its security controls. (CryptoWatchdog may earn a commission if you open an account via our Kraken link; it does not affect our assessment.)

8. Artificial lock-up periods

Lock-ups are often used to prevent a bank run. By forcing depositors to leave funds in place for, say, 90 days, operators ensure they have enough liquidity to honour the small early withdrawals that build trust, while delaying the day of reckoning.

Some legitimate staking and DeFi protocols use lock-ups for sound technical reasons, so context matters. But a mandatory lock-up on an "AI trading" product deserves scrutiny. If the algorithm is truly profitable and liquid, you should be able to access your principal reasonably quickly. Long, inflexible delays often signal that the treasury is thin and the operators are buying time.

This is one area where AI-washing overlaps with otherwise legitimate structures, so weigh it alongside the other flags rather than in isolation. Genuine tokenised assets, for instance, can carry redemption terms for real reasons, as we explain in our overview of real-world asset tokenisation. The difference is whether the terms are clearly justified or simply impose friction on getting your money back.

9. Selective withdrawal success

Scammers frequently allow small withdrawals to work flawlessly at first. This is deliberate. A successful 50-dollar withdrawal convinces you the platform is real and encourages a far larger deposit later.

The trouble begins when you try to take out a meaningful sum. Suddenly there is "technical maintenance," a surprise "KYC audit," or a demand for a "tax" or "unlock fee" before your funds can be released. None of those gates ever truly opens.

If you reach this point, do not send more money to "unlock" your balance. Paying again only marks you as a target for follow-up fraud. This same pattern fuels a secondary wave of recovery scams that prey on people who have already lost funds, so be sceptical of anyone who contacts you out of the blue promising to retrieve your money for an upfront fee.

What regulators and researchers are saying

You do not have to take a watchdog's word for it. Several official and well-documented sources now describe AI-washing directly:

• SEC, March 2024. The Commission brought its first AI-washing cases against two advisers for misstating their use of AI, a move covered by the SEC's own press release and by legal analysts at Harvard Law School's corporate governance forum.
• FTC, September 2024. Operation AI Comply brought multiple enforcement actions against firms making deceptive AI claims, signalling that exaggerated AI marketing is itself an enforcement target.
• FCA, ongoing. The UK regulator's crypto investment scams guidance warns that fraudsters fake prices and returns and that consumers may have no access to the Financial Ombudsman Service or compensation scheme if things go wrong.
• Chainalysis, 2026 report. Researchers have reported that generative AI is making investment scams more convincing and scalable, with crypto scam losses reaching record levels as automation increases both speed and credibility. Figures vary between reporting cycles, so treat the headline numbers as estimates from the firm rather than settled fact.

The common thread is that AI claims are now enforceable representations. A platform that lies about its AI is not just exaggerating; in several jurisdictions it may be breaking the law.

Practical checks to perform before you deposit

Run through these steps before committing any capital. None requires technical expertise, and together they catch the majority of AI-washed schemes.

1. Search for trouble. Look up the project name with the words "scam," "Ponzi," and "review" on independent forums and search engines. Read the negative results, not just the testimonials.
2. Check the domain age. A company claiming years of track record but running on a domain registered last month is misrepresenting itself.
3. Verify regulatory status. Most AI-washed platforms claim to be "regulated" but cannot give a licence number you can confirm on a regulator's public register. If you cannot verify it, assume it is false.
4. Test the technology claim. Ask, in plain English, what data the model uses and how it decides. Vagueness is the answer.
5. Inspect the API request. Never grant withdrawal permissions to any bot or platform.
6. Confirm custody. Establish whether you control your funds at all times, or whether you are handing them to a third party with no protection.

Common mistakes to avoid

• Letting FOMO override judgement. Scammers borrow the credibility of real AI firms to manufacture urgency and make you feel you are getting in early. Urgency is a sales tactic, not a fact about the opportunity.
• Clicking links in unsolicited messages. Direct messages promising AI trading often lead to wallet-drainer sites that empty your balance the moment you connect. Navigate to sites yourself rather than through links sent to you.
• Reusing withdrawal-enabled API keys. Audit and revoke old permissions periodically using a reputable token-approval tool, and keep long-term holdings off connected platforms entirely.
• Trusting screenshots. Treat any gain you cannot independently verify as marketing, not evidence.

The verdict on AI and crypto

The genuine intersection of AI and blockchain is a real and active field. But the overwhelming majority of retail-facing "AI yield" products marketed with guaranteed returns are not innovation; they are old frauds wearing a new costume. They weaponise the complexity of AI as a shield against the basic question every investor should ask: how, exactly, is the money made?

If a platform trips more than one or two flags on this list, the prudent move is to step away. The downside is total loss of capital, and no plausible upside justifies that risk on an unverifiable promise. Protecting your money here is rarely about being clever; it is about being patient enough to verify before you act.

Frequently asked questions

Is all AI in crypto a scam? No. Machine learning is used legitimately for on-chain analytics, fraud detection, market research, and risk modelling. The problem is not AI itself but false or exaggerated claims about it, which is what regulators call AI-washing. Judge each project on verifiable evidence rather than on whether it uses the term "AI."

What is the single biggest red flag? Promises of fixed, guaranteed, or "risk-free" returns with no losing days. Real trading involves volatility and drawdowns. A perfectly smooth upward curve is the clearest sign you are looking at a Ponzi structure dressed up as an AI bot.

Have regulators actually punished AI-washing? Yes. The SEC brought its first AI-washing enforcement actions in March 2024 against two investment advisers, and the FTC launched Operation AI Comply later that year against firms making deceptive AI claims. AI claims are now treated as representations that can be enforced against.

A platform let me withdraw a small amount, so it must be real, right? Not necessarily. Allowing small withdrawals to build trust before blocking larger ones is a well-documented scam tactic. Consistent withdrawals at any size matter far more than one successful small test.

Should I give an AI trading bot my exchange API keys? Only ever with trade-only permissions, and never with withdrawal rights enabled. A genuine bot needs to place trades, not move your money off the exchange. For long-term holdings, keep them in self-custody so no bot can reach them.

How can I check if a project's team is genuine? Search the founders' names alongside their claimed credentials and employers, look for a track record that predates the project, and be cautious of anonymous teams, stock-photo profiles, or accounts created very recently. AI now makes fake personas easy to generate, so verify endorsements come from real, identifiable people.

Where can I report a suspected AI-washing scam? In the UK, report to the FCA and to Action Fraud. In the US, you can report to the SEC, the FTC, and the FBI's Internet Crime Complaint Center. Reporting helps regulators build cases and warn others, even if recovery is uncertain.

Final checklist

• Are the returns realistic for a volatile market, with honest losing periods?
• Can I see a live, independently verified trade history rather than screenshots?
• Is the team public, accountable, and backed by a real track record?
• Does the business model depend on recruiting new members?
• Can the team explain the technology in plain language?
• Do I keep custody of my funds, and have I refused withdrawal API permissions?

If you cannot answer those confidently in the project's favour, you are most likely looking at an AI-washed product. Stay sceptical, verify before you deposit, and remember that walking away costs nothing.

Related reading

• Best crypto exchange UK 2026
• Self-custody versus custodial wallets
• Real-world asset tokenisation: gold, silver, real estate
• Kraken review
• CryptoMine Pro scam warning
• YieldMax AI scam warning