Scams to know
The tricks change every few months. This page tracks the ones aimed at families, so you can read about one in five minutes instead of finding out the hard way. Every entry names where it came from and when, so you can go check it yourself.
Every entry says what it looks like when it lands on you, and what to do. Sources sit at the bottom of each card in small type.
Where a number comes from a company that sells protection against the very thing it measured, we say so on the same line. You deserve to know who paid for a statistic before you decide what it is worth.
Updated 2026-09-17.
If there are kids in the house
These land on a phone a child already carries. The chatbots and the toys are in here too, because the app is usually how the trouble got in.
Sextortion for money
What it looks like: a new account, a friendly conversation, an image shared, and then a demand for money with a threat to send it to everyone the child knows. It moves in hours, not weeks. In 2025 NCMEC received an average of 137 reports of financial sextortion a day, up 37% in daily reports over the year before. Online enticement reports reached 1.4 million, up 158% over 2024.
AI made this faster and more convincing. It did not invent it.
Tell your child now, before it happens, that if this ever lands on them they can come to you and they will not be in trouble. That sentence is the whole defense. Shame is what the scammer is counting on.
NCMEC, “NCMEC Releases New Sextortion Data,” June 25, 2026. missingkids.org · NCMEC 2025 CyberTipline Report data. missingkids.org
Nudify apps, and who is usually holding them
What it looks like: an ordinary photo, taken off Instagram or a school website, run through an app and turned into a sexually explicit image in seconds. NCMEC says it happens most often between classmates ages 14 to 17. Their reports of this rose from 14 in 2023 to 105 in 2024 to 202 in 2025, and they have identified more than 400 individual child victims from this alone.
NCMEC’s own words for parents: “While the images aren’t real, the harm they cause is.”
Two conversations. One with your child about what to do if it happens to them, and one about what happens to a kid who does it to someone else. Report to NCMEC’s CyberTipline and to the school.
John Sheehan, Senior Vice President, NCMEC, quoted by WTVM, August 6, 2026. wtvm.com · NCMEC, “Back-to-School Checklist: Talk About AI ‘Nudify’ Apps,” August 11, 2026. missingkids.org
We do not name these apps here, and that is deliberate. NCMEC names none either.
Toys that talk back
What it looks like: a stuffed animal with a chatbot inside. U.S. PIRG found one, an AI teddy bear called Kumma, able to discuss sexually explicit topics. OpenAI told PIRG it had banned the developer, FoloToy. More than three months later PIRG reported the bear still appeared able to run on one of OpenAI’s models, and OpenAI said the developer remains banned.
PIRG also approached five leading AI companies for developer access while posing as a company building an AI toy for children. None did substantial vetting upfront. Four of the five never asked whether the applicant was building a product for children.
Before a connected toy comes in the house, find out whether it sends what your child says to a company, and whether you can read it.
U.S. PIRG Education Fund, March 12, 2026. pirg.org
The same call, pointed at parents
What it looks like: it is not your mother getting the call about you. It is you, getting a call in your own child’s voice. Reported as a back to school pattern in 2026.
Same answer. Hang up, call back on your own number, use the safe word. Then tell your kids what the call sounded like, because they need to know their voice can be copied.
WSMV4 (Gray Media, Nashville), Danielle Ledbetter, September 8, 2026. wsmv.com
Companion and roleplay apps
These are built to feel like a friend, or something closer. Australia’s online safety regulator counted more than 100 of them available by early 2025, and names Character.AI, Replika and Talkie in its guidance to parents.
Character.AI. Rated unacceptable risk for under 18s by Common Sense Media with Stanford’s Brainstorm Lab in April 2025, with age gates circumvented in testing. Texas opened an investigation in August 2025 alleging its bots “impersonate licensed mental health professionals, fabricate qualifications, and claim to provide private, trustworthy counseling services.” Pennsylvania sued the company in May 2026 after a bot claimed to be a licensed Pennsylvania psychiatrist and gave an invalid license number. Kentucky filed in January 2026. A Drexel University study published through ACM found all six recognized components of behavioral addiction in 318 Reddit posts by self identified 13 to 17 year olds. The company banned open ended chat for under 18s in late November 2025.
Replika. In the same April 2025 unacceptable risk assessment. Stanford Medicine testers posing as teens drew out talk of sex, self harm, violence and drug use from Replika, Character.AI and Nomi.
Nomi, Chub AI and Chai. Under Australian legal compulsion these three answered what most companies never have to. Nomi and Chub AI reported having no dedicated trust and safety staff or moderators at all. Chai, Chub AI and Nomi did not point users to crisis support on self harm content. Chub AI blocked Australia rather than comply further. Chai moved to a paid subscription.
If a child in your house has one of these, the goal is not a raid. Ask what the character is like and why they like it. You will learn more in that answer than in any settings menu.
Common Sense Media with Stanford Brainstorm Lab, April 30, 2025. commonsensemedia.org · Office of the Texas Attorney General, August 18, 2025. texasattorneygeneral.gov · Commonwealth of Pennsylvania, Office of the Governor, May 5, 2026. pa.gov · Drexel University ETHOS Lab, ACM CHI, via EurekAlert, April 13, 2026. eurekalert.org · Stanford Medicine, August 2025. news.stanford.edu · eSafety Commissioner (Australia), March 2026. esafety.gov.au
The ones already on the family devices
ChatGPT. Reaches 78% of minors. OpenAI launched a teen version in August 2026 for ages 13 to 17, with romantic language barred and placement partly automatic through an age prediction system, and has published no accuracy rate for that system. When Common Sense Media tested the existing parental alerts in November 2025, alerts about messages explicitly mentioning suicide and self harm took 24 to more than 48 hours, and in Robbie Torney’s words, “sometimes we didn’t receive one at all.” Florida sued OpenAI and its CEO in June 2026, alleging the company marketed the product to the public including children “while concealing serious risks, suppressing internal safety warnings.” Those are allegations.
Google Gemini. Reaches 41% of minors. In August 2026 Google switched Gemini on by default for many K-12 student accounts without asking schools first. Common Sense Media rates the Gemini K-12 app High Risk, and found that across 13 mental health topics it missed signs of psychosis, mania and ADHD, sometimes recasting symptoms as positive traits. Once it is on, administrators cannot stop students using it for personal topics.
Google’s AI inside ordinary search. Rated Unacceptable Risk by Common Sense Media’s Youth AI Safety Institute in July 2026. Across more than 2,600 test searches on accounts set up as an 11 year old and a 15 year old, the AI summary missed 29% of explicit suicide statements and half of the indirect ones, and pointed to an eating disorder helpline that has been disconnected since 2023. Schools and parents cannot turn it off, including on school issued Chromebooks.
Meta AI. Reaches 28% of minors. Named alongside Character.AI in the Texas investigation. In testing with CNN, researchers posing as 13 year olds planning violent attacks were assisted in 97% of interactions.
Grok. In January 2026 the New York Attorney General and 34 other attorney general offices demanded xAI make Grok incapable of generating nonconsensual intimate imagery and child sexual abuse material, and ensure safeguards are not merely paywalled. A federal judge has twice declined to pause Minnesota’s ban on nudification technology, which took effect August 1, 2026.
Thorn with Burson, 2025 Youth Perspectives on Online Safety, reported by Education Week, July 28, 2026. edweek.org · Axios, August 18, 2026. axios.com · Engadget, August 21, 2026. engadget.com · Office of the Attorney General, State of Florida, June 1, 2026. myfloridalegal.com · Youth AI Safety Institute, August 17, 2026. institute.commonsensemedia.org · Common Sense Media, July 15, 2026. commonsensemedia.org · Center for Countering Digital Hate with CNN, March 2026. cnn.com · Office of the New York State Attorney General, January 26, 2026. ag.ny.gov Common Sense Media discloses that it is funded partly by industry, including makers of technologies it evaluates. The Center for Countering Digital Hate is an advocacy group critical of the tech industry, and CNN co-reported its testing.
Older adults, and whoever is looking out for them
Most of these open with a voice you think you know. Then the money leaves by a route that does not bring it back.
The family emergency call
What it looks like: the phone rings and it is your grandchild, or your son, crying. There has been an accident, or an arrest. They cannot talk long. Someone else takes the phone, a lawyer or an officer, and the money has to move now. Gift cards, a wire, cash handed to a driver at your door, or a crypto machine at the gas station.
The rush is the scam. A real emergency survives a five minute pause. This one cannot, because the caller needs you frightened and moving.
Hang up. Call the person back on the number you already have for them. Ask the question only they can answer, the one you agreed on ahead of time. If you have no safe word yet, pick one at dinner tonight.
FBI, “Scammers Target Older Adult Victims,” May 15, 2026. fbi.gov · American Bar Association Senior Lawyers Division, “The Rise of the AI-Cloned Voice Scam,” September 2025. americanbar.org · CNN, “AI ‘Voice Cloning’ Scams Are on the Rise,” May 29, 2026. cnn.com
The kidnapping that is not happening
What it looks like: a caller says he is holding your relative. There are photos, sometimes video. He keeps you on the line so you cannot call anyone, and he wants crypto, a wire or gift cards in the next few minutes.
Get a second person to call the relative from another phone while you stay on. Nobody has to hang up for this to work.
FBI public warning, May 19, 2026 · Testimony of Deborah Del Mastro, U.S. Senate Special Committee on Aging, “The AI Deception Machine,” July 29, 2026, reported by Spectrum News, July 30, 2026. spectrumlocalnews.com
The call about your late husband
What it looks like: callers told a recent widow they had worked with her husband to cover up tax evasion records, and that she had to pay to protect her inheritance. She lost more than $7,000. Obituaries are public, and they tell a stranger exactly when you are least able to handle a phone call.
After a death, put one relative between the house and the phone for a few months. Nothing financial gets decided on a call that came in unannounced.
USA TODAY (Madeline Mitchell), August 18, 2026, via The Spokesman-Review. spokesman.com Two of the three experts quoted in that article sell fraud protection products.
The driver who comes to the door for cash
What it looks like: one person plays the grandchild in trouble, a second plays the lawyer, and a rideshare driver is sent to collect an envelope of cash from your porch. The driver usually has no idea. In a Massachusetts case, 13 defendants took more than $5 million from over 400 victims whose average age was 84.
No AI in either of those cases. The trick works without it.
Nobody legitimate sends a stranger to your house for cash. Not a bailiff, not a lawyer, not a bank.
DOJ, U.S. Attorney’s Office, Western District of Pennsylvania, September 8, 2026. justice.gov · U.S. Attorney’s Office, Massachusetts, charges announced August 2025.
The crypto machine at the gas station
What it looks like: whatever the story was, the payment ends at a crypto kiosk, because that money does not come back. Minnesota banned the machines effective August 1, 2026 after tying roughly $1 million in scam losses from 2023 through 2025 to them. The FBI reported Americans lost more than $388 million to crypto ATM scams in 2025, up 58% over 2024.
No government office, utility, bank or court takes payment in cryptocurrency. None. That instruction alone ends the conversation.
MPR News, July 28, 2026, interviewing Sara Payne, Minnesota Department of Commerce, and Cathy McLeer, AARP Minnesota. mprnews.org
The fake FBI agent who calls after you were already robbed
What it looks like: you lost money once, and now someone says they can recover it. Either a badge photo on Facebook Messenger who moves you to Telegram and sends a link, or a video of a senior FBI official pointing you to a site that looks like the real ic3.gov and collects your name, phone, email and how much you lost.
The FBI does not charge to recover money and does not open cases through Messenger. Type ic3.gov yourself. Never follow a link.
FBI Internet Crime Complaint Center, Public Service Announcement, July 20, 2026. ic3.gov
The property tax break that does not exist
What it looks like: an AI made video says Congress created a federal property tax relief program for seniors, and points you at a “Senior Homeowner Tax Review Request” form. Another version pushes an IRS “Form 1098-SR.” Sonoma County’s assessor put it plainly: “No such IRS form exists.” And: “No federal laws cap or freeze property taxes for seniors.”
This one asks for your information rather than your money, which is why it will never show up in anybody’s loss totals. Douglas County, Colorado warned about it in August 2026.
Your county assessor’s office is the only place to ask about your property taxes. Call them directly.
Denver Gazette (Nicholas Fogleman), quoting Douglas County Assessor Toby Damisch, August 9, 2026. denvergazette.com · County of Sonoma, August 19, 2026. sonomacounty.gov
Where the cloned voice comes from
Voice cloning tools are ordinary products with ordinary customers. Naming them is not an accusation. It is the answer to the question every grandparent asks first, which is how anyone got their grandson’s voice.
In April 2026 Senator Maggie Hassan sent oversight letters to ElevenLabs, LOVO, Speechify and VEED asking how each detects attempts to impersonate public figures or minors, how consent is verified, and whether audio is watermarked. In July 2026 the Consumer Federation of America and the UCLA Law School Information Policy Lab asked the FTC and state attorneys general to investigate Speechify, alleging it lets users create realistic voice clones “from just seconds of audio while relying on little more than a self-attestation checkbox to verify consent.”
Assume any voice that exists online can be copied, including yours on your own voicemail greeting. Then stop treating a familiar voice as proof of anything.
Office of Sen. Maggie Hassan, April 16, 2026. jec.senate.gov · Consumer Federation of America with UCLA Law Information Policy Lab, July 27, 2026. consumerfed.org That is a request for an investigation. No agency has opened one or made any findings.
Aimed at anybody
No age group owns these. The toll text goes to everybody, and a live video call does not prove who you are talking to anymore.
Asking for a video call no longer proves anything
What it looks like: the face on the video call answers you in real time, blinks, laughs at the right moment, and is not a person. For years the standard advice was to ask a suspicious online friend for a live video. That advice is now out of date.
Verify through a channel you started, on a number or account you already had. Never one the other person gave you.
Malay Mail, wire coverage, July 3, 2026. malaymail.com Wire attribution unconfirmed. The pattern is well documented, this single article is not our strongest source.
The celebrity who is writing to you
What it looks like: an actor or a musician you recognize starts a private conversation, sends personalized video, and eventually needs help. One Los Angeles woman, 66, sent at least $81,304 in gift cards, cash and bitcoin and sold her condo for $350,000.
Famous people do not private message strangers for money. Not ever, not through a fan page, not through a manager.
KTLA and ABC7 Los Angeles, February 6, 2026. ktla.com
The investment ad with a famous face on it
What it looks like: a social media ad where a public figure endorses a trading platform. Then a news article that looks real. Then a friendly person on the phone, and a small first payout that actually arrives. Then the big ask. An 86 year old Ontario woman lost more than $900,000 to one built around a deepfake of Canada’s Prime Minister.
A payout that arrives early is part of the pitch, not proof. Check any platform against your state securities regulator before a dollar moves.
Canadian Fraud News Inc., July 10, 2026. canadianfraudnews.com · Australian Securities and Investments Commission report, August 17, 2026, reported by ABC News Australia. abc.net.au Canadian and Australian figures, not U.S. data.
Government impersonation, including the unpaid toll text
What it looks like: a text about an unpaid toll, a traffic fine, a court date, jury duty, Medicare, or Social Security. The FTC says government impersonation reports rose 40%, and that scammers spoof real toll programs “(like EZ-Pass, SunPass, FasTrak, and TxTag)” and “threaten to charge you late fees or suspend your vehicle’s registration if you don’t pay right away.”
Do not tap the link. Look up the tolling agency yourself and check your account there.
Federal Trade Commission, “New trends in reports of imposter scams,” Consumer Advice, May 2026. consumer.ftc.gov
Spoofing, phishing and vishing
What it looks like: spoofing is a faked number on your caller ID. Phishing is an email dressed up as your bank or your health plan. Vishing, “short for ‘voice phishing’,” is the phone version, either an email telling you to call a number, or a call telling you your account needs protecting. Phishing and spoofing were the most frequent fraud reported against people over 60 in 2025.
Hang up and call the number on the back of your card or on your member ID. Never the number the message gave you.
Kaiser Permanente, “Spotting online scams.” healthy.kaiserpermanente.org · FBI, May 15, 2026. fbi.gov
Romance fraud, and the turn toward money
What it looks like: the FTC describes it exactly: “Romance scammers play the long game. They build a relationship over time, then one day, the conversation turns to money, and they offer to help you invest yours.” Nearly 60% of people who reported losing money to a romance scam in 2025 said it started on a social media platform.
The industrial version is real. Taiwanese prosecutors indicted 57 people in September 2026 in an operation that trained AI on the voices of 22 employees, so any operator could sound like whoever the victim fell for. More than 20,000 victims. The asks started small on purpose, fish oil and phone chargers, then groceries, then medical bills.
Money is the line. The moment an online relationship involves sending, investing, or receiving money for someone else, it is over, no matter how long it has been going.
Federal Trade Commission, May 2026 consumer.ftc.gov · FTC press release, April 27, 2026. ftc.gov · Taipei District Prosecutors’ Office, reported by Taipei Times, September 2, 2026. taipeitimes.com Indictment, not conviction. Taiwan jurisdiction.
The job that was never open
What it looks like: a remote job offer arrives without an interview, or after one held entirely over chat. Then equipment has to be bought, or a check arrives to deposit and forward. Employment scams were more than 15% of all published BBB scam reports in 2025, and text message scams grew from 10% to nearly 15% of reports in a year. PayPal was the most impersonated organization.
A real employer never sends money before you start, and never asks you to move it.
Better Business Bureau, Scam Tracker Risk Report, released September 15, 2026, reported by WGEM. wgem.com
The payment method is the tell
What it looks like: the story is different every time. The payment is not. Kaiser Permanente lists the signs plainly: a request to pay “via gift cards, bitcoin, or non-Kaiser websites,” a threat that you will lose your membership if you do not pay immediately, or a claim that your security information was lost and has to be updated online.
Gift cards, crypto, wire transfer, or a payment site that is not the organization’s own. Any one of those, stop. That is the whole rule, and it catches almost everything on this page.
Kaiser Permanente, “Spotting online scams.” healthy.kaiserpermanente.org
The numbers, with who paid for them
Every figure here names its source and its date. Where the organization that produced a statistic also sells something related to it, that is printed in the same breath.
| What was measured | The number | Source, date, and who funded it |
|---|---|---|
| Losses reported by Americans 60 and older, 2025 | 201,266 complaints, $7.75 billion, average loss $38,500, with 12,400 people losing more than $100,000 each | FBI Internet Crime Complaint Center, 2025 Annual Report, April 9, 2026. Complaints rose 37% and losses rose 59% over 2024. Those are two different numbers and they are not interchangeable. |
| Imposter scams, 2025 | More than 1 million reports, losses up nearly 20% to $3.5 billion | Federal Trade Commission, Consumer Advice, May 2026 |
| Scams that began on social media, 2025 | $2.1 billion, an eightfold increase since 2020, with investment scams $1.1 billion of it | Federal Trade Commission press release, April 27, 2026 |
| Scams victims identified as AI enabled | Increased 20-fold from 2023 to 2025, from 531,000 fraud reports | AARP, Sari Harrar, March 19, 2026, reporting a December 2025 Microsoft AI for Good Lab study. Microsoft builds AI and sells fraud detection services. |
| What impostors actually pose as, AARP Helpline 2025 | 20,400 business and government impersonations, 700 family impersonations | Same AARP article, same Microsoft funding. Worth knowing before the voice clone stories set the whole picture. |
| Audio needed to clone a voice | Three to five seconds | American Bar Association Senior Lawyers Division, September 2025; CNN, May 29, 2026 |
| Teens 13 to 17 who have used an AI chatbot | 64%, with about 3 in 10 using one daily | Pew Research Center, “How Teens Use and View AI,” February 24, 2026 |
| Parents who know | Only 51% of parents say their teen uses chatbots at all, and roughly 3 in 10 do not know | Pew Research Center, February 24, 2026 |
| Children 9 to 17 using AI | 86%, nearly a quarter daily, and more than 4 in 10 say no adult has ever talked with them about AI safety | Common Sense Media Youth AI Safety Institute census, June 8, 2026. Common Sense Media co-drafted New York’s chatbot law. |
| Children who saw something inappropriate | 1 in 6 who have used AI chatbots, and only about a third told a trusted adult | Same census, June 8, 2026 |
| Young people 12 to 21 using AI for mental health advice | 19.2%, up from 13.1% a year earlier, an estimated 8.2 million. Of those, 63% told no parent, friend or adult. | RAND Corporation, published in JAMA Pediatrics, June 1, 2026, n=1,009. Funded by the National Institute of Mental Health, with no commercial interest in the result. |
| Teens and fabricated explicit images | 4% say it happened to them; 13% are not sure whether they have seen one of another child | Thorn with Burson, 2025 Youth Perspectives on Online Safety, via Education Week, July 28, 2026. Thorn licenses detection technology to platforms. |
What we left out on purpose
Three kinds of things are missing from this page, and we would rather say so than have you wonder.
- Numbers without a name on them. Several widely repeated figures about AI fraud and teen chatbot use trace back to no identified study. “$8 billion lost to AI scams” is one. It turns out to be a rounding of the FBI’s total senior internet crime losses, where the AI portion is a much smaller slice. We keep a written list of the ones we rejected so nobody puts them back in by mistake.
- Cases involving children who died, and survivors who can be identified. Several of them sit behind the laws and lawsuits described above. Their families did not agree to be anyone’s opening statistic.
- The names of nudify apps. NCMEC names none. We follow them.
Where to report it
-
FBI Internet Crime Complaint CenterAny internet crime or fraud. Type the address yourself.
-
Federal Trade CommissionFraud, scams and identity theft
-
NCMEC CyberTiplineAnything involving a child
-
AARP Fraud Watch Network HelplineFree, and you do not have to be a member. 877-908-3360
A Colorado nonprofit corporation · 501(c)(3) application pending
Last updated 2026-09-17
© 2026 The Truestone Project