Key Takeaways
- Researchers flagged more than 18,000 suspected scam messages circulating across 3,300 Spanish-language public WhatsApp groups over an eight-month stretch, with at least 30% of the phone numbers tied to U.S. codes, according to the Digital Democracy Institute of the Americas.
- Impostor scams — the category that covers AI-cloned-voice calls — cost people in the U.S. a reported $3.5 billion in 2025, and FTC data show older adults tend to lose far larger individual amounts than younger victims.
- Consumer-protection regulators say a few seconds of audio, often lifted from a public social media post, is now enough to fake a familiar voice — which is why official guidance centers on a rehearsed family "safe word" and hanging up to call back a number you already trust.
A Platform Built on Trust Becomes a Target
For millions of Latino households in the U.S., WhatsApp isn't just an app — it's the phone line connecting them to parents, siblings, and childhood friends still living abroad. That reliance is well documented: DDIA's review of Pew Research Center survey data found that as of early 2024, 54% of U.S. Latino adults used WhatsApp regularly, a higher share than any other racial or ethnic group tracked, and up eight points from 2021. That same closeness is exactly what scammers are learning to exploit, because a message or call that looks like it comes from inside the family circle rarely gets the scrutiny a stranger's message would.
What the Research Actually Shows
Between January and September 2025, researchers at DDIA and the social-listening firm Palver combed through thousands of public, Spanish-language WhatsApp groups and identified more than 18,000 individual messages suspected of pushing scams or fraud, spanning fake giveaways, bogus job offers, sham legal and health services, and dubious investment pitches, according to the group's three-part investigation. Some of that content spread widely enough that Meta itself marked it with a "frequently forwarded" label. To measure how the problem has evolved, the researchers also reviewed 139 Spanish-language fact-checks published since 2017 and found that 2025 alone produced 43 of them in ten months — roughly one a week, and the fastest pace on record.
It's worth being precise here: this dataset tracks scam and disinformation messages broadly, not specifically AI-generated audio. The overlap between the two — voice-cloning fraud riding the same Spanish-language group-chat infrastructure DDIA studied — is a documented pattern in the wider fraud landscape, but it hasn't yet been measured with the same rigor inside these specific WhatsApp groups.
How an AI-Cloned Call Actually Works
The mechanics are consistent across the cases regulators and security researchers have documented. A scammer pulls a short audio sample — sometimes just a few seconds — from a public video, voicemail greeting, or voice note, then feeds it into inexpensive voice-cloning software. The resulting call typically leans on urgency and secrecy at once: a supposed accident, arrest, or medical emergency, paired with pressure not to tell anyone else in the family. A related tactic, described in a recent report on messaging-app fraud, pairs voice spoofing with fake job recruiters who request "processing fees" and QR codes disguised as delivery or account-verification prompts. The end goal in nearly every version is the same: get the victim to wire money, send cryptocurrency, or hand over gift-card codes before there's time to double-check.
Why Older Adults Absorb the Biggest Losses
The FTC's newest fraud figures don't break impostor-scam losses out by ethnicity, so no agency has published a verified dollar figure specific to elderly Latino victims. But the age pattern alone is striking. Nationally, the FTC reported that people lost a combined $3.5 billion to impostor scams in 2025 — the fifth straight year that category topped fraud complaints — out of roughly $15.9 billion lost to fraud overall, a 27% jump from 2024. In the most recent year the agency has broken down by age, 2024, adults 60 and older accounted for $1.6 billion, or 68%, of that group's total losses through cases involving $100,000 or more — evidence that when older victims get scammed, they tend to lose far more than younger ones.
How to Verify a Call Before You Act
Federal regulators have converged on a short list of defensive habits. Per FTC guidance on family emergency scams, the first rule is simple: don't trust the voice, no matter how convincing. Hang up, then call the person back on a number already saved in your phone rather than one texted or read aloud during the suspicious call. If that person can't be reached directly, try another relative or close friend before sending anything. Agencies also recommend agreeing on a private family code word in advance — something never posted online — that a real relative could supply on request and a cloned voice could not.
What Families Can Do Inside the Group Chat Itself
Because much of this content spreads through public WhatsApp groups rather than one-on-one calls, individual vigilance only goes so far. WhatsApp's own "frequently forwarded" label — the double-arrow icon DDIA's researchers used to help flag high-risk content — is a built-in warning sign worth pausing on before resharing. The app's native reporting and blocking tools let any member flag a suspicious message directly to Meta, and group administrators can remove repeat offenders rather than simply muting a thread. Before forwarding anything urgent-sounding, a quick check against Spanish-language fact-checking outlets that participate in the Google Fact Check API — the same network DDIA drew on for its research — can confirm whether a claim has already been debunked.
Regulators Are Watching the Same Platforms
Concern about Spanish-language fraud and disinformation on encrypted messaging apps has reached Capitol Hill. Sen. Ben Ray Luján and colleagues, including Sens. Bob Menendez and Alex Padilla, sent a formal inquiry pressing WhatsApp, Telegram, and Signal on what they're doing to curb the spread of Spanish-language mis- and disinformation, arguing that Latino communities' heavy reliance on these apps makes the platforms' response a matter of public urgency. Separately, the FTC has leaned on its Impersonation Rule to bring a string of enforcement actions against companies posing as government agencies or businesses, though none so far have targeted the Spanish-language WhatsApp-group ecosystem specifically.
The Bottom Line
The technology behind AI voice cloning has gotten cheap and fast enough that a handful of seconds of audio is now a real liability. Combined with how central WhatsApp and Telegram already are to Latino family life, that makes a rehearsed verification habit — not suspicion of every call, but a quick, calm check before sending money — the most practical defense available right now.
