@@@
How to spot an AI grifter: Starring Mel Robbins
Encryptress
Some of you know Mel Robbins from The Let Them Theory, the 2024 book that turned two words into a whole personality and sold a few million copies. She’s spent years building her brand as the no-nonsense friend who tells you the thing you really needed to hear. What you might not know is that in May she aimed all of that trust at 12 million people and told them to upload their bank statements to a chatbot.
“You cannot be left behind,” she touted to her followers in a video few realized was sponsored. The post was paid for by Microsoft, tagged #copilotpartner tucked way at the very bottom of the caption where nobody reads. Pinned in the comments was a copy-paste prompt for using Microsoft Copilot to “take control of your money” and to make that script actually work, it asked people to hand the chatbot their bank statements, debt statements, bills, and income info. So the advice, from a woman getting paid to give it, was to feed the single most sensitive pile of data you own straight into a consumer AI tool.
Security people of course clocked it instantly. The thing about typing into a free or consumer-tier chatbot is a lot of the time you are not in a private room (drop a comment if you’re confused what I mean by this). Depending on the product and your settings, what you paste can get stored, fed into training the next version of the model, or read by an actual human reviewer. Account numbers, balances, income, who you owe - all of it. Once that stuff is sitting on somebody else’s servers you don’t have any say over where it goes, who see it, or what next quarter’s quietly updated terms let them do with it. Dr. Jen Gunter, a physician with a large following of her own, said it straight: do not upload your financial information to this tool. A cybersecurity engineer named Jas Wong made the practical version of the point which is: if you’re going to do this at all, the post needed a giant flashing warning to black out your Social Security and account numbers first.
Microsoft says its AI isn’t built to replace professional advice, and that Copilot lets you control what it remembers or opt out of the retention thing entirely. Robbins, once the internet came for her, said she could’ve framed the example more clearly and that they updated the prompt language after the fact. Fine. None of that lived in the post 12 million people actually saw and acted on. The caveats showed up after the backlash, which is basically always the whole problem. The reach went out first and the fine print came in later, once it didn’t matter anymore.
We’ve actually seen this before
Take a step back and consider this playbook. A trusted famous person stands in front of an enormous audience and says get in NOW, don’t be the one who gets left behind. They are being paid to speak on something they know very little about and the product is riskier than the pitch lets on. By the time everybody works that out, the damage has been done. This isn’t a Mel Robbins original, AI is just the latest thing getting promoted off the back of it.
We watched this same machine run on crypto. The SEC charged Kim Kardashian for hyping a token called EthereumMax without properly disclosing she’d been paid $250,000 to do it. She settled for $1.26 million and got barred from promoting crypto for three years. Back in 2018 both Floyd Mayweather and DJ Khaled paid fines for pushing crypto to their followers without admitting they were compensated. Trusted name, paid placement, risky product, audience left holding the empty bag and the response is lol why the hell would you trust DJ Khaled (honestly though… sorry).
AI is running a very similar play, except there’s no securities law in the way. Nobody is getting fined for telling you to dump your tax returns into a chatbot. No regulator is drawing the line for you, so the only thing standing between the pitch and your data is whether you can recognize the grift.
Let’s talk about it!
How to spot an AI grifter
The tell is almost never the product itself. It’s the language wrapped around it - similar to phishing emails actually. These pitches reach for the same tired handful of moves every single time, and once you learn to hear them you genuinely can’t un-hear them:
None of this means AI is inherently predatory, or that being curious about it makes you a target. It means the safe version of this advice looks nothing like the sponsored one. Ask a chatbot to explain a confusing line on your bank statement, have it define a term, run a what-if on a made-up number. Do NOT go handing it the real documents with your actual name and account numbers all over them and when somebody with millions of followers and a brand-partner tag tells you you’re behind and you’d better hurry, stop right there and take a breath. The urgency IS the ad.
Read more: SEC.gov | Cybernews | SheKnows
“You cannot be left behind,” she touted to her followers in a video few realized was sponsored. The post was paid for by Microsoft, tagged #copilotpartner tucked way at the very bottom of the caption where nobody reads. Pinned in the comments was a copy-paste prompt for using Microsoft Copilot to “take control of your money” and to make that script actually work, it asked people to hand the chatbot their bank statements, debt statements, bills, and income info. So the advice, from a woman getting paid to give it, was to feed the single most sensitive pile of data you own straight into a consumer AI tool.
Security people of course clocked it instantly. The thing about typing into a free or consumer-tier chatbot is a lot of the time you are not in a private room (drop a comment if you’re confused what I mean by this). Depending on the product and your settings, what you paste can get stored, fed into training the next version of the model, or read by an actual human reviewer. Account numbers, balances, income, who you owe - all of it. Once that stuff is sitting on somebody else’s servers you don’t have any say over where it goes, who see it, or what next quarter’s quietly updated terms let them do with it. Dr. Jen Gunter, a physician with a large following of her own, said it straight: do not upload your financial information to this tool. A cybersecurity engineer named Jas Wong made the practical version of the point which is: if you’re going to do this at all, the post needed a giant flashing warning to black out your Social Security and account numbers first.
Microsoft says its AI isn’t built to replace professional advice, and that Copilot lets you control what it remembers or opt out of the retention thing entirely. Robbins, once the internet came for her, said she could’ve framed the example more clearly and that they updated the prompt language after the fact. Fine. None of that lived in the post 12 million people actually saw and acted on. The caveats showed up after the backlash, which is basically always the whole problem. The reach went out first and the fine print came in later, once it didn’t matter anymore.
We’ve actually seen this before
Take a step back and consider this playbook. A trusted famous person stands in front of an enormous audience and says get in NOW, don’t be the one who gets left behind. They are being paid to speak on something they know very little about and the product is riskier than the pitch lets on. By the time everybody works that out, the damage has been done. This isn’t a Mel Robbins original, AI is just the latest thing getting promoted off the back of it.
We watched this same machine run on crypto. The SEC charged Kim Kardashian for hyping a token called EthereumMax without properly disclosing she’d been paid $250,000 to do it. She settled for $1.26 million and got barred from promoting crypto for three years. Back in 2018 both Floyd Mayweather and DJ Khaled paid fines for pushing crypto to their followers without admitting they were compensated. Trusted name, paid placement, risky product, audience left holding the empty bag and the response is lol why the hell would you trust DJ Khaled (honestly though… sorry).
AI is running a very similar play, except there’s no securities law in the way. Nobody is getting fined for telling you to dump your tax returns into a chatbot. No regulator is drawing the line for you, so the only thing standing between the pitch and your data is whether you can recognize the grift.
Let’s talk about it!
How to spot an AI grifter
The tell is almost never the product itself. It’s the language wrapped around it - similar to phishing emails actually. These pitches reach for the same tired handful of moves every single time, and once you learn to hear them you genuinely can’t un-hear them:
- “Don’t get left behind.” Manufactured urgency. The only job that sentence has is to get you acting before you’re thinking. Anything actually worth using does not expire because you took a week to understand it first.
- “Everyone is already doing this.” Peer pressure. Robbins leaned on a study about women using AI less than men to build this exact you’re-falling-behind panic.
- “This isn’t financial advice, but...” A liability dodge that lets someone hand you financial advice while technically swearing they didn’t. Kardashian’s crypto post used almost this precise phrasing.
- “I wish I’d had this back when I was struggling.” The personal-story. It reframes a paid ad as a warm little tip from a friend who’s been right where you are.
- A tiny hashtag at the bottom #ad or #copilotpartner buried under three lines of caption is technically disclosure and functionally invisible. If you have to go hunting for the part where they admit they got paid, congrats, you just found the most important part of the whole post.
None of this means AI is inherently predatory, or that being curious about it makes you a target. It means the safe version of this advice looks nothing like the sponsored one. Ask a chatbot to explain a confusing line on your bank statement, have it define a term, run a what-if on a made-up number. Do NOT go handing it the real documents with your actual name and account numbers all over them and when somebody with millions of followers and a brand-partner tag tells you you’re behind and you’d better hurry, stop right there and take a breath. The urgency IS the ad.
Read more: SEC.gov | Cybernews | SheKnows