It’s no surprise that online scammers have taken to generative AI – the success of these scams, after all, is often a numbers game. The more people they can attempt to reach, and the more ways they can attempt to reach them, the better the odds are that someone will eventually fall for the grift. Generative AI makes it easy for scammers to spin stories at scale, and, as Vectra AI reports, scams using AI grew by 1,210% in 2025, with the FBI reporting $893,346,472 worth of AI scam-related losses.
So what does this mean for the Ponzi scheme in 2026? It’s a lot worse than it was before.
What is a Ponzi Scheme?
Before we dive into Ponzi schemes in the post-AI era, let’s lay a foundation – what is a Ponzi scheme? Unlike the very sudden and very recent rise of large language models and generative AI, Ponzi schemes have been around since the 1920s (or even earlier, depending on how strict we’re getting with the definition). They’re named for Charles Ponzi, who made millions by promising victims returns on a nonexistent investment opportunity – in this case, a big return on postal reply coupons – while using money from new investors to pay out early investors.
Who's Calling Me?
Search any phone number to learn more about the owner!
How It Works
AI or no, the basic Ponzi scheme operates on the same structure as it has since before the Great Depression. For all 100-plus years of its sketchy history, the Ponzi scheme has typically followed this format:
The grifter promises “investors” big returns with little risk…
…then pays early investors with money from later investors.
This creates the illusion of great ROI – easy money (which is almost always a myth) – and thus generates lots of positive word-of-mouth.
Based on that word-of-mouth, the Ponzi perpetrator continues to collect money from new investors.
Inevitably, the flow of new money slows. Once withdrawals outpace new deposits, the scheme collapses, and most investors lose their money.
Heard of Bernie Madoff’s 2008 grift to collect $64.8 billion worth of blue-chip stocks and options based on completely false records of past profits? That historical investment fraud is a prime example of a classic Ponzi scheme.
The New Wave: Ponzi 3.0
Ponzi schemes are one of the oldest tricks in the fraudster’s book, but what does a Ponzi scheme look like today? Well, speaking of Bernie Madoff, Harvard Business School Professor Eugene Soltes tells Harvard Business Review that today’s Ponzi schemes “make Bernie Madoff's fraud look trivial.” And the differentiating factor is – you guessed it – artificial intelligence.
Today, fraudsters run the same racket as their predecessors, but use easily accessible, cheap, and convincing-at-a-glance AI-generated assets like slick websites and fake video testimonials. They may even use AI to create completely bogus profiles of previous investors (because what is a Ponzi scheme without some flashy zhuzhing?).
Alongside these assets, some modern-day Ponzi schemes are fully automated, carried out entirely by AI chatbots (autonomous agents based on large language models). In addition to facilitating more scams than we’ve ever encountered before, essentially flooding the zone with bait for victims, AI-automated Ponzi schemes leverage some unique advantages in their execution. According to the aforementioned source at Harvard, these advantages often boil down to three key takeaways:
AI bots – nowadays employed by virtually every major company on the internet – can collude to fix prices in real time.
Personalized Ponzi schemes. GenAI is algorithmic; it vacuums up tons of data and regurgitates it in the shape it “thinks” (or more accurately, attempts to predict) that humans want in a given situation. Chatbot-powered Ponzis can craft schemes on the fly that are specifically tailored to a given victim’s interests, spending habits, or past behaviors.
Professor Soltes predicts more extreme measures in the future: “Perhaps it’s a bot that wants to maximize an investment. So, it creates a smart contract to cause physical harm to a competitor’s leadership or physical infrastructure to increase its value, which would automatically be paid out to whoever completes the contract.” The combination of algorithmic specificity and a lack of human empathy is bad news for potential Ponzi victims.
You don’t need to be a Harvard professor pontificating about a Skynet future to see that the AI-assisted Ponzi scheme has already arrived in a very big way. Of the nearly $900 million lost to AI-powered scams in 2025, about $632 million of those losses were suffered at the hands of investment fraud involving artificial intelligence to target victims.
Casual Investment Fraud – Now With AI
Not all AI-related investment frauds are deepfake-powered, chatbot-automated Ponzi schemes. Because of the massive hype around AI, many grifters are simply using promises of the technology – whether it’s a hot new AI “innovation” that doesn’t exist or a groundbreaking development in the tech itself – to fraudulently attract investors.
In these cases, the money collected is very real, but the AI-flavored promises that attracted said money are not. In other cases, fraudsters make fabricated AI claims about online investment platforms themselves. This has become so rampant, in fact, that the SEC and its partners have issued an official warning about unregistered or unlicensed investment platforms claiming to use AI.
How to Spot an AI-Powered Ponzi Scheme
Because AI-driven Ponzi schemes personalize themselves to the victim, and can use a wide variety of assets – from chatbots to fake social media profiles to vibe-coded websites – consistent red flags can be hard to pin down. Oftentimes, your best bet is identifying the AI itself; whether it’s a Ponzi scheme or not, any “investment opportunity” offered up by a chatbot with a bogus site and testimonies is probably not something you want to give your cash to.
Social profiles that are blank, brand new, have zero followers, or seem to be populated exclusively with botted followers are a red flag. Likewise, check for AI-generated profile pics or weird usernames, like those mostly composed of random numbers or letters.
In terms of images and videos, waxy, overly smooth skin, implausible anatomy, oversaturation, improper interaction with physical objects, and garbled, nonsense text are dead giveaways. Look for AI brand watermarks, like the diamond-shaped Gemini logo, in the corner of any pics, too.
Be wary of commonly overused chatbot phrases, like “certainly,” “I understand your concern,” “great question,” and “absolutely.” Remember that most chatbots will try to placate you and tell you you’re right, no matter what you say, before contradicting you.
Overly quick responses – like an essay-length text or DM that appears instantly – given at all times of the day can also be a red flag.
If you’ve got a hunch, try using some tricks to reveal an AI chatbot. Ask them to switch to French for two sentences before resuming English, ask them the time and date in their time zone, or request that they describe the view out their window. Ask them lots of follow-up questions, specifically calling back to details you’ve mentioned. While these tests aren’t foolproof, strange or slightly off answers should certainly add to your suspicion.
As genAI gets harder to identify, don’t hesitate to use free online AI detection tools for images and text.
And of course, the red flags for regular, human-made Ponzi schemes still apply. Keep an eye out for suspicious giveaways like no-risk, high-return investments that seem too good to be true, secretive or hard-to-understand strategies, and account statement issues or other consistent paperwork hiccups. Before you invest with anyone, always make sure any investment firms you deal with are fully licensed, and that investments are registered with the SEC.
Outsmart AI
One thing Ponzi and Madoff had over AI chatbots is that they were both real human beings, for as crooked as they were. And while we can’t throw AI chatbots in prison (yet) like the lifetime sentences both historic schemers got, that doesn’t mean we’re out of options.
A smart investor always does their research, but if you’re unsure whether a potential partner is the real deal or an AI chatbot, try a quick Spokeo People Search. Just type in a name, phone number or email address, and we’ll instantly run it against more than 60 billion open-source records, 1.7 billion social media profiles, and much more. The results will clue you in on whether your contact is who they say they are, or if they’re even a person at all – and that sort of intelligence is very real.
As a freelance writer, small business owner, and consultant with more than a decade of experience, Dan has been fortunate enough to collaborate with leading brands including Microsoft, Fortune, Verizon, Discover, Office Depot, The Motley Fool, and more. He currently resides in Dallas, TX.
