Crypto scams are the most documented practical risk around digital assets, and generative AI is making them more convincing. In 2025, consumers reported USD 15.9 billion in total fraud losses to the US Federal Trade Commission, and investment scams alone accounted for USD 7.9 billion of that. An Ontario Securities Commission experiment found that participants put 22% more money into AI-enhanced fake investments than into conventional ones. This page is general information, not investment advice.
Key takeaways
- Investment scams were the costliest fraud category reported to the FTC in 2025, at USD 7.9 billion out of USD 15.9 billion.
- Cryptocurrency payments lack the fraud protection that credit cards offer, and Canada's anti-fraud agency (the CAFC) says official bodies never demand crypto payment.
- The OSC tested scam materials polished with generative AI and found they drew 22% more simulated investment than conventional scam materials.
- In the same experiment, a simulated browser plug-in that flagged high-risk offers cut simulated investment in fraudulent offers by 31%.
- Reported loss figures are self-reported and incomplete, so they show direction, not exact totals.
What reported losses show
The FTC's June 2022 data spotlight counted more than 46,000 people who reported losing over USD 1 billion in cryptocurrency to scams since the start of 2021. That was about one in four dollars lost to fraud in the period. Investment scams made up USD 575 million of the total, and nearly half of those who named a starting point said the scam began with an ad, post, or message on social media.
The picture has grown larger since. In testimony delivered on March 25, 2026, the FTC reported 3 million fraud reports and USD 15.9 billion in losses for 2025, with investment scams on top at USD 7.9 billion. The FTC's testimony does not break out the crypto share for 2025, so no crypto-only figure is given here.
Canadian numbers point the same way. The CAFC fraud trends page for January to June 2026 lists investment fraud as the costliest category, at about CAD 139.6 million in reported losses, and says the pitches spread through social media, online ads, messaging apps, and impersonation. That figure covers all investment fraud, including crypto-related schemes, not crypto alone. Every number in this section is self-reported to a government agency, and many victims never report.
Why crypto payments suit fraudsters
Most losses start with a pitch, not a hack. The CAFC describes a pattern it calls romance-investment fraud, sometimes nicknamed pig butchering: a scammer builds trust on a dating app or social media, then steers the victim toward a fake trading platform. Victims are told to send crypto to an address the fraudster controls, and the site may display convincing but invented balances to encourage more deposits.
Two properties of crypto payments help the scammer. A transfer is rarely reversible once confirmed, and the agency notes that crypto lacks the fraud protection that credit cards provide. The same agency says no government body will ever ask for payment in cryptocurrency, which makes any such demand a reliable sign of a scam.
What generative AI changes
The OSC and the consultancy BIT Canada published a study in 2024 on AI and retail investing scams. Their literature and environmental scan concluded that generative AI is turbocharging familiar investment scams by increasing their reach, efficiency, and effectiveness. It also enables new forms, such as deepfakes and voice cloning, and fake offers that borrow the appeal of the phrase "AI-enhanced investing."
The study also ran a randomized controlled trial with over 2,000 Canadian participants in a simulated social media feed. Offers included ETFs, cryptocurrencies, and advisory services, and mixed legitimate offers with conventional scams and AI-enhanced scams. Participants invested 22% more in the AI-enhanced scams than in the conventional ones. These were hypothetical decisions in an online simulation, so the result shows relative persuasiveness, not real-world loss rates.
Where AI helps on the defense side
The same trial tested two protections. An inoculation step, which gave short guidance on scam awareness before exposure, reduced investment in fraudulent offers by 10%. A simulated browser plug-in that flagged potentially high-risk offers reduced it by 31%. The authors conclude that protections built into platforms and systems, followed by individual-level education, are both needed.
The plug-in in the experiment was a simulation, not a product, and no real-world detection rate can be read from it. It does show the logic of defense tools: attach a visible warning at the moment of decision. Any real tool also has to weigh false alarms, because a warning on a legitimate offer erodes trust in the tool.
Red flags and where to check them
The table below turns the CAFC's published warning signs into a quick reference. Treat it as a sample layout for a check, not a guarantee that an offer without these signs is safe.
| Warning sign | Why it matters | Where to verify |
|---|---|---|
| Unsolicited offer, even from a friend or relative | Accounts get hijacked and trust gets borrowed | Contact the person through a separate channel |
| Returns described as high or guaranteed | Only scammers guarantee profits, per the FTC | Provincial or territorial securities regulator, or the SEC in the United States |
| Pressure to act quickly | Urgency prevents checking | Wait, and research the platform name plus the word "fraud" |
| Payment demanded in crypto | Hard to reverse and no card-style protection | Bank, local police, and the CAFC or the FTC fraud reporting service |
| Trading site shows large balances but blocks withdrawals | Balances can be fabricated | Stop sending money and report |
Information, not advice
This page describes documented scam patterns and research on detection. It does not assess any token, platform, or strategy, and it makes no statement about whether any crypto asset will gain or lose value. Digital assets can lose most of their value quickly, and losses in a scam are usually permanent. Anyone considering a financial decision can speak with a registered adviser, and readers in Canada can check a firm's registration with their provincial securities regulator before sending money.
FAQ
Are crypto losses covered the way card fraud is?
Generally not. The CAFC states that crypto lacks the fraud protection credit cards offer, and the FTC advises that payments in crypto typically cannot be reversed.
Can AI reliably spot a crypto scam?
Not on its own. The OSC experiment showed that an automated warning can cut simulated losses by 31%, which leaves most of the risk in place. Human awareness and regulator checks remain part of the defense.
Why do scammers use AI?
The OSC reports that generative tools raise the reach, speed, and polish of scam materials, and enable deepfakes and voice cloning that were not possible before.
Where should a suspected scam be reported?
Canadian residents can use the CAFC and their local police, and contact their bank and provincial securities regulator. US residents can report to the FTC fraud reporting service.
Related reading
- How AI and cryptocurrency differ as technologies
- Documented concerns about Bitcoin
- What AI does around crypto markets
- Where AI and crypto projects overlap
More in AI in finance.
Last verified: October 2026
Sources
- Ontario Securities Commission and BIT Canada, "Artificial Intelligence and Retail Investing: Scams and Effective Countermeasures," 2024: source page
- CAFC, "Investment" fraud page: CAFC investment page
- CAFC, "Fraud Trends," January to June 2026, page modified August 25, 2026
- US Federal Trade Commission, testimony before the Joint Economic Committee on fraud, March 25, 2026
- US Federal Trade Commission, "Reported crypto scam losses since 2021 top $1 billion, says FTC Data Spotlight," June 2022