AI trading scams: what the regulators have documented

AI-branded fraud is the largest documented consumer harm in this market — larger than model error, larger than platform failure. ASIC coordinated the removal of 11,964 phishing and investment scam websites during 2025, against 6,270 in the preceding twelve months, a 90% increase, and more than 25,000 sites since 2023. "AI" is the dominant marketing term in that material because it supplies a plausible-sounding answer to the one question a fraudulent operator cannot otherwise handle: how are these returns produced? The verification that defeats most of it takes two minutes in a public register.

Published 27 August 2026 · Updated 28 August 2026 · AI Trading Book Editorial · Reading time about 14 minutes

The documented scale

Australia produces the most complete public numbers of the three markets we cover, so the figures below are Australian unless stated. They understate the true position everywhere, because takedown counts measure enforcement activity rather than the number of operations running, and loss figures capture only reported cases.

The investment share of total losses is worth holding in mind: of AUD 2.18 billion in reported scam losses for 2025, AUD 837.7 million was investment fraud — a little over a third of all reported scam harm in the country, concentrated in one product category.

Why fraud reaches for the word "AI"

Every investment fraud has to answer one question: where does the money come from? Historically the answers were forex arbitrage, offshore property or a proprietary trading floor. AI is a better answer than any of them, for three reasons that have nothing to do with technology.

It is unfalsifiable to a non-specialist. A claim that a neural network finds patterns humans cannot see cannot be checked by the target, and asking for detail invites a plausible refusal about proprietary methods. It borrows genuine credibility, because real institutions genuinely do use machine learning, so the category is not absurd on its face. And it explains consistency — the hardest thing for a fraud to justify is why returns never vary, and "the algorithm doesn't have emotions" is a ready-made answer.

ASIC Commissioner Alan Kirkland has described the pattern directly, noting that scammers often claim to "leverage the latest AI technology to make money with minimal effort." The two halves of that phrase are the offer: sophistication you cannot evaluate, and effort you do not have to make.

Documented fabricated brands

These are not real firms behaving badly. They are invented names with no licensed entity behind them, built to be searched for after a victim sees an advertisement.

Note the naming pattern. Both borrow authority from real technology history — quantum computing, and the IBM chess system that beat Kasparov in 1997. A name that sounds like it belongs to a research programme is doing work that a name like "FastProfit Ltd" cannot.

Both brands also outlived their takedowns. Removing a domain does not remove the operation; the same material reappears under a new name, which is why the takedown count keeps rising while the pattern stays constant.

The recurring funnel

The sequence below is assembled from regulator warnings and consistent complaint patterns. It is not a single documented case study but a composite of the stages that recur. The important structural point is that the early stages are designed to feel legitimate, and the extraction happens only after trust is established.

Why the small withdrawal matters most

The single most effective element in the sequence is allowing an early withdrawal of a small amount. It costs the operator a few hundred dollars and converts a suspicious prospect into a confident one, because it appears to answer the one test a careful person applies: can I actually get my money out?

It does not answer that test. It demonstrates that the operator chose to return a small sum at a moment when doing so was profitable. Treat an early successful withdrawal as a neutral fact, not as verification — and note that the request to deposit substantially more almost always follows it.

Deepfake and celebrity-impersonation endorsements

Fabricated endorsement is now the leading distribution method. ASIC reported AUD 7.4 million in losses from celebrity-impersonation scams in media release 26-195MR of 17 August 2026, alongside the removal of more than 19,400 scams.

The economics changed. Producing convincing video or audio of a public figure endorsing a trading platform used to require resources; it now requires very little. That inverts the old heuristic, in which a video of a known person was strong evidence of legitimacy. The current position is closer to the opposite: a video endorsement of a specific trading platform by a celebrity, politician or central banker is far more likely to be fabricated than real, because legitimate financial firms in these jurisdictions face strict rules on how investments may be promoted and rarely use that format for a specific product.

Fabricated news coverage works the same way. Cloned layouts of real publications, complete with mastheads and bylines, are cheap to produce and are typically served only to people arriving from the advertisement rather than being reachable from the publication's own site.

Red flags and what each one actually indicates

How to check a licence in each market

This is the check that resolves most cases, and it works because it does not depend on judging how professional something looks. Type the register address yourself; do not follow a link supplied by the platform, since fake register pages exist.

Four checks that convert a register lookup into a real verification:

1. Search the entity name, not the brand. Trading names can be borrowed; the legal entity is what is registered.
2. Compare the contact details on the register with the ones you were given. A mismatch in phone number, email domain or address is the signature of a cloned-firm scam, and it is the fastest way to detect one.
3. Confirm the permissions match the offer. A licence to deal is not a licence to advise, and a licence in another country is not a licence here.
4. Check the regulator's warning lists before depositing. ASIC, the FMA and the FCA all publish lists of firms they have warned about.

Availability is not authorisation

An application that installs and works in your country proves only that it has not been geo-blocked. An offshore operation without local authorisation is not bound by conduct rules, is not required to belong to a dispute resolution scheme, holds your money under rules you have not read, and cannot be reached by your regulator.

This is the point at which the fraud question blurs into the regulation question. Not every unlicensed offshore platform is a scam — but the protections described on our UK, Australian and New Zealand regulation pages simply do not apply to it, and if something goes wrong the practical difference between an unlicensed offshore firm and a fraudulent one is small.

If you have already paid

Act on the payment first; everything else can follow. Speed matters most for card and bank transfers, where a reversal is sometimes possible if raised quickly.

1. Contact your bank or card issuer immediately and ask them to attempt to stop or reverse the payment.
2. Stop all further payments. Any demand for a fee, tax or fine to release your funds is part of the fraud.
3. Preserve everything — messages, screenshots of the dashboard, transaction records, names, phone numbers, wallet addresses and website URLs.
4. Report it. Australia: ASIC and Scamwatch, run by the National Anti-Scam Centre. New Zealand: the FMA. United Kingdom: the FCA and Action Fraud.
5. Secure your accounts. If you installed remote-access software or shared credentials, change passwords from a different device and enable two-factor authentication.
6. Expect a recovery approach and refuse it. Victim lists circulate; an unsolicited offer to recover your money for an upfront fee is a second fraud.

Report even if you lost nothing. Takedown coordination — the activity that produced the 11,964 figure — depends on reports of sites that have not yet claimed a victim.

This is not an argument that all AI trading tools are fraudulent

Established platforms with published pricing, identifiable corporate entities and verifiable regulatory status exist; several are documented on our platforms page with their tariffs and limits. The distinguishing features are mundane: a named legal entity, a published price list rather than a bespoke offer, a licence you can find without being sent a link, and explicit acknowledgement that losses are possible.

The honest framing is that legitimate tools and fraudulent ones make different promises. A real platform sells analysis, execution or automation and tells you that you may lose money. A fraud sells an outcome. Our evidence page covers what the legitimate category actually delivers, which is considerably less than the marketing on either side suggests.

What we could not establish

• A breakdown of investment scam losses attributable specifically to AI-branded fraud. The AUD 837.7 million figure covers investment scams as a category. Data not found.
• Equivalent takedown statistics for New Zealand or the United Kingdom. No comparable published series was identified, so the figures on this page are Australian and should not be read as global. Data not found.
• A recovery rate for funds lost to these operations. No official measure of how much is recovered was identified. Data not found.

Key takeaways

• Fraud, not model error, is the documented harm. 11,964 sites removed in one year, up 90%, with more than 25,000 since 2023.
• "AI" is doing narrative work, not technical work. It answers the question of where returns come from in a way that cannot be checked.
• Any guaranteed return is decisive. The guarantee is the red flag; the percentage is decoration.
• Video endorsement now argues against legitimacy. Fabrication is cheap and documented at AUD 7.4 million in losses.
• A working small withdrawal proves nothing. It is the most reliable element of the funnel.
• The register check is the whole defence. Type the address yourself, search the legal entity, and compare the contact details.

Frequently asked questions

How common are AI trading scams?

Common and growing. ASIC coordinated the removal of 11,964 phishing and investment scam websites during 2025, against 6,270 in the preceding twelve-month period — a 90% increase — plus more than 1,100 social media investment scam advertisements. More than 25,000 sites have been removed since 2023.

Why do scams use the word AI?

Because it supplies a plausible explanation for impossible returns. ASIC Commissioner Alan Kirkland has noted that scammers often claim to leverage the latest AI technology to make money with minimal effort. AI lets a fraudulent operator answer the obvious question — how are these returns generated — without disclosing a strategy.

What fake AI trading brands has ASIC documented?

ASIC has named "Quantum AI" in media release 24-180MR and "Deep Blue" in 26-063MR. The Deep Blue material advertised returns of 85% and 374%. These are fabricated brands with no licensed entity behind them, not real firms behaving badly.

How much money is lost to investment scams in Australia?

The National Anti-Scam Centre reported total scam losses of AUD 2.18 billion for 2025, of which AUD 837.7 million was investment scams. The 2024 total was AUD 2.03 billion, a 25.9% fall on the prior year. Reported losses understate the real figure because many victims never report.

Are celebrity endorsements of AI trading platforms real?

Overwhelmingly not. ASIC reported AUD 7.4 million in losses from celebrity-impersonation scams in media release 26-195MR of 17 August 2026, alongside the removal of more than 19,400 scams. Video and audio of public figures endorsing trading platforms is now cheap to fabricate and should be treated as evidence of fraud, not endorsement.

How do I check whether a trading platform is licensed?

Use the national register directly, typed in yourself rather than through a link the platform gave you. Australia: ASIC Connect professional registers. New Zealand: the Financial Service Providers Register plus the FMA's licensed provider information. United Kingdom: the FCA Financial Services Register. Confirm the licence covers the specific service offered.

A platform showed me a licence number. Is that enough?

No. Cloning a real firm's licence number is a documented pattern. Look the number up on the register yourself, then compare the contact details on the register with the ones you were given. If the phone number, email domain or address differ, you are dealing with an impersonator, not the licensed firm.

Why did my small withdrawal work?

Because it is part of the method. Allowing an early small withdrawal establishes credibility and typically precedes a request for a much larger deposit. It costs the operator little and reliably increases the eventual loss. An early successful withdrawal is not evidence that a platform is genuine.

What returns should make me suspicious?

Any specific promised return, at any level. Fraudulent material documented by ASIC advertised 85% and 374%. But modest promises are equally diagnostic: a guaranteed 1% per week is a guaranteed 67% per year, which no legitimate provider can promise. The red flag is the guarantee, not the size of the number.

Someone offered to recover my lost funds for a fee. Is that legitimate?

Almost certainly not. Recovery fraud targets people who have already lost money, often using victim lists from the original scam, and asks for an upfront fee or new personal details. Legitimate recovery goes through your bank, the police and the regulator, and does not require an advance payment to a private operator.

What should I do if I have already paid money to a scam?

Contact your bank or card issuer immediately to attempt to stop or reverse the payment, stop all further payments, keep every message and transaction record, and report to the relevant national body. Do not make additional payments to release funds — that request is part of the fraud, not a route out of it.

Does a professional-looking website mean a platform is real?

No. Fraudulent operations run polished websites, live dashboards showing rising balances, and responsive support staff. The balance shown in a dashboard is a number in the operator's database, not a custodial record. Presentation quality carries no information about legitimacy.

Are all AI trading platforms scams?

No. Established platforms with published pricing, named corporate entities and verifiable regulatory status exist and are covered on our platforms page. The distinction is not whether AI is mentioned but whether the entity is identifiable, licensed for what it offers, and honest about the possibility of loss.

Where do I report an AI trading scam?

In Australia, to ASIC and to Scamwatch run by the National Anti-Scam Centre. In New Zealand, to the FMA, which maintains a warning list. In the United Kingdom, to the FCA and to Action Fraud. Report even if you did not lose money — takedown coordination depends on reports.

Sources

• ASIC — 26-063MR, 8 April 2026 — 11,964 phishing and investment scam websites removed during 2025 against 6,270 in the preceding twelve-month period, a 90% increase; more than 1,100 social media investment scam advertisements; more than 25,000 sites since 2023; the fabricated "Deep Blue" brand advertising 85% and 374% returns; Commissioner Alan Kirkland on scammers claiming to use the latest AI technology.
• ASIC — 24-180MR — the fabricated "Quantum AI" brand.
• ASIC — 26-195MR, 17 August 2026 — AUD 7.4 million in losses from celebrity-impersonation scams; more than 19,400 scams removed.
• National Anti-Scam Centre — reported scam losses: AUD 2.18 billion total for 2025 including AUD 837.7 million in investment scams; AUD 2.03 billion for 2024, a 25.9% decrease.
• ASIC Connect professional registers; Financial Service Providers Register and FMA licensed provider information; FCA Financial Services Register — verification routes described in Table 4.

Informational research only. Nothing on this page is personal financial, legal or investment advice. Naming a brand as documented in a regulator publication reflects that publication, not an independent finding by us. If you believe you have been targeted, contact your bank and the relevant national regulator; if you are in immediate financial distress, contact your bank's hardship team, which exists for this purpose.