Scam Watch

Le truffe più diffuse — riconoscile prima di cascarci

Schede rapide con segnali d'allarme, azioni immediate, fonti e link per verificare un caso sospetto.

Grandparent emergency scam — bail / hospital cash courier

Caller (sometimes AI voice clone of grandchild from social media samples) tells grandparent 'I've been arrested / in an accident, don't tell mom and...

Medicare 'new card' scam — health insurance ID harvesting

Caller posing as Medicare or insurer says a 'new Medicare card with chip / DNA testing kit / back brace' is being issued and asks for the Medicare...

Military romance scam — fake deployed soldier impersonation

Scammer creates profile on Match, OurTime, Facebook using stolen photos of real US military member (often general or special forces), claims...

Sweepstakes / lottery winner scam — Publishers Clearing House impostor

Senior receives call/letter: 'Congratulations, you've won $2.5M from Publishers Clearing House / Mega Millions / Costco'. To claim the prize, victim...

Beware of fake inheritance notifications asking for upfront fees

Scammers send convincing messages claiming you are the sole heir of a distant relative or unknown wealthy stranger. They promise millions and ask you...

Advance-Fee Scams: How Fraudsters Reinvents Old Traps with Geopolitical Hooks

...

Meta AI Support Bot Handed Instagram Accounts to Hackers - How to Protect Your Account

Meta's AI powered support chatbot was tricked by attackers into changing the email addresses on Instagram accounts, effectively handing over account...

FBI: Americans lost nearly $900 million to AI-powered scams in 2025

The FBI's 2025 Internet Crime Report reveals $893 million in AI related scam losses across 22,364 complaints, with voice cloning, deepfake videos,...

AI generated fake seeds sold as impossible plants on eBay, Amazon, and Etsy

Scammers are using AI image generators to sell seeds for plants that do not exist, listing them on eBay, Amazon, and Etsy. Listings promote flowers...

AI-facilitated fraud surging: 22,000+ IC3 complaints, USD 893M+ losses (2025) — voice clones, deepfakes, AI chatbots

In 2025 the FBI IC3 received more than 22,000 complaints citing AI involvement, with losses exceeding USD 893 million. Criminals use generative AI...

How to spot AI-generated images used in scams: fake pets, dating profiles, and donation appeals

Scammers now use AI generated images to fabricate emotional stories and trick people into sending money or sharing personal data. From fake lost pet...

Detecting AI-generated images used in scams - NCSC Switzerland weekly review Week 28

Scammers increasingly use AI tools to create deceptively realistic images for phishing emails, fake shop websites, social media profiles, and...

Deepfake KYC bypass — synthetic ID + liveness deepfake

Sumsub 2024 reported a 10x rise in deepfake based KYC attacks on banks / exchanges / brokers. Attacker generates fake ID document (passport /...

AI-generated fake employment reference + recommendation letter

Job applicant (or fraud ring like DPRK IT workers) generates fake reference letters via GPT 4 / Claude — perfect business English, plausible names,...

AI-generated invoice with pixel-perfect vendor logo + signature

Attacker uses Midjourney / SDXL / GPT Vision to clone a vendor's invoice template — exact logo, watermark, formatting, even a fake authorising...

AI-generated profile photos on dating apps + LinkedIn (StyleGAN / Midjourney)

Scammers populate Tinder / Bumble / LinkedIn / Instagram with photos generated by StyleGAN3, Midjourney, FLUX. Used for pig butchering, romance, fake...

AI Brand Phishing: How ChatGPT, Claude, and DeepSeek Lures Steal Credentials and Cards

Threat actors are exploiting the AI hype wave by impersonating popular platforms like ChatGPT, Claude, DeepSeek, and Microsoft Copilot in phishing...

Instagram Account Hijack via Meta AI Support Bot: How It Works and How to Stay Safe

Hackers are exploiting Meta's new AI customer support chatbot to hijack Instagram accounts, including high profile handles like the Obama White House...

LLM-written BEC email — perfect grammar across languages, no telltale typos

Threat actors use ChatGPT / Claude / local LLMs to write Business Email Compromise lures in flawless English, Italian, German, French, Mandarin etc....

AI essay-mill scam — students pay for 'human-written' work that's GPT-generated

Essay mill websites charge students $20 200 per assignment claiming 'human writers'. In reality, ChatGPT / Claude generate the work in minutes, mill...

AI-generated fake court / police document for extortion or fee fraud

Attackers use multimodal LLMs to generate convincing court summons, police citations, prosecutor letters — perfect typography, official looking...

AI-generated fake news network for ad-revenue + influence operations

NewsGuard (2024) tracked 1000+ AI generated 'news' websites — full editorial content written by GPT/Claude, headshots StyleGAN, RSS scraped +...

Full-AI romance — no human, end-to-end LLM chatbot relationship

Latest evolution: the scammer is a fully automated chatbot pipeline (LLM for chat + voice TTS for occasional voice notes + AI image gen for 'selfies...

AI-generated 'finfluencer' promoting bogus trading platform

Fully synthetic 'finance influencer' (no real human) on Instagram/TikTok/X promotes a 'proprietary AI trading bot' with screenshots of 1000% returns....

Multi-party deepfake Zoom call impersonating CFO + execs (Arup $25M HK case Jan 2024)

Attackers harvested LinkedIn/YouTube/webinar footage of Arup execs, trained real time face swap + voice clone models, then invited a Hong Kong...

Statecraft deepfake — fake former Ukrainian FM Kuleba Zoom (Sen. Cardin Sept 2024)

An apparent nation state actor (Russia/China/Iran suspected) impersonated former Ukrainian FM Dmytro Kuleba in a live Zoom with US Senator Ben...

Celebrity-endorsed deepfake investment video (Musk / Trump / Bocelli / Schumacher)

Hijacked YouTube channels (sometimes 1M subs) livestream deepfake 'Elon Musk' / 'Trump' announcing 'send 1 BTC / ETH to this address, receive double...

Diplomatic / NGO Zoom-meeting setup with deepfake of known interlocutor

Threat actor scrapes public footage of a known ex official / NGO leader / scholar the target would reasonably take a meeting with. Schedules a Zoom...

Deepfake dirigeant — fausse visio Teams avec PDG / DAF

Mode opératoire identique à Arup / WPP, mais ciblé sur entreprises françaises (CAC40, ETI): visio Teams/Zoom où le 'PDG' et le 'DAF' sont des...

Deepfake forensic tells — iris reflection asymmetry, blink anomaly, neck-edge artifacts

Reality Defender + Hive + Truepic + DARPA SemaFor identify multiple deterministic tells in deepfaked video. Useful for analysts + journalists...