Millions of Americans hand over personal information every day. They share their data with insurance companies, banks, investment apps, and other services they trust.
And that’s exactly why cybercriminals target and impersonate those services.
This week, an insurance provider disclosed a breach reportedly affecting nearly 7 million people’s driver’s license numbers, while a California journalist shared how a convincing fake Robinhood text ultimately cost her more than $70,000.
Here’s what happened, why these scams work, and what you can do to protect yourself This Week in Scams.
Nearly 7 Million Driver’s License Numbers Exposed in Insurance Data Breach
One of the largest U.S. data breaches of the year has exposed sensitive information belonging to 6.9 million people.
According to reporting from TechCrunch, insurance provider AssuranceAmerica confirmed that hackers accessed customer information after compromising an employee account. The company says the stolen data includes names, contact information, driver’s license numbers, insurance policy details, vehicle information, and claims data.
While the company has not said exactly how the employee’s credentials were compromised, it noted that the attackers targeted an employee account before accessing company systems.
Why driver’s license numbers matter
Unlike a password, you can’t simply change your driver’s license number.
Combined with your name, address, phone number, or other information from previous breaches, driver’s license numbers can be used by criminals to:
Open fraudulent accounts
Impersonate victims during identity verification
Make phishing scams more convincing
Support broader identity theft schemes
This is also part of a larger trend. In recent months, multiple breaches have exposed government-issued identity documents as more organizations collect IDs for identity verification and age-check requirements.
If you receive a notice that your information was involved in a breach, monitor your financial accounts closely, consider placing a fraud alert or credit freeze, and remain cautious of unexpected emails, texts, or phone calls referencing your insurance or driver’s license information.
Unfortunately, scammers will reach out saying they’re trying to “help” secure your stolen information, only to try and steal more personal data from you.
How McAfee Can Help Before, During, and After a Data Breach
Before a breach
Personal Data Cleanup helps reduce your digital footprint by removing your personal information from many data broker sites, limiting what scammers can easily find about you.
During a breach
Identity Monitoringalerts you if your personal information appears on the dark web or in known data leaks, helping you respond faster if your information is exposed.
After a breach
Scam Detector helps identify suspicious texts, emails, and links that often follow major breaches, while Web Protection helps block malicious websites designed to steal additional information or credentials.
Fake Robinhood Text Scam Costs Former News Anchor More Than $70,000
Even people who report on scams can become victims.
A former California television news anchor recently shared how she lost more than $70,000 after receiving what appeared to be a legitimate text message claiming there was suspicious activity on her Robinhood investment account.
The message instructed her to call a phone number for assistance. Once connected, the caller posed as Robinhood support before transferring her to a fake “fraud department.”
Believing she was protecting her investments from hackers, she was convinced to move her money into what she thought was a secure account. Instead, it went directly to scammers.
She later contacted Robinhood through the official app, but by then the money had already been transferred.
Why investment scams are becoming more convincing
Investment scams rely on urgency, authority, and impersonation rather than obvious phishing emails.
Rather than asking targets to “invest” immediately, many scams begin by convincing people that their existing account is under attack and immediate action is needed.
At McAfee, we’ve also seen scammers impersonate Robinhood, Charles Schwab, cryptocurrency platforms, and other investment services through fraudulent text messages and malicious links promising AI-powered investing, exclusive bonuses, or unusually high returns.
Whether the message claims your account has been compromised or promises incredible profits, the goal is often the same: get you to click, call, or transfer money before you have time to verify what’s happening.
Investment Safety Checklist
Before responding to any message about your investments:
Never call the phone number provided in a text message or email. Instead, contact your financial institution using the number listed in its official app or website.
Slow down when someone creates urgency. Claims that your account is being hacked or frozen are designed to make you act before you think.
Be skeptical of guaranteed returns or AI-powered investment opportunities. Promises of extraordinary profits are a common hallmark of investment fraud.
Verify alerts through your account directly. If you receive a suspicious notification, log in through the official app, not a link in the message.
How McAfee Can Help
With McAfee+, multiple layers work together before any damage is done:
Scam Detectorflags suspicious texts, emails, links, QR codes, and even deepfake videos before you engage
Secure VPNkeeps your data private, especially on public Wi-Fi
Web Protection helps block risky sites, even if you do accidentally click
Password Manager doesn’t just help you make unique, strong passwords, it keeps them stored and organized for you
Americans submitted more than 1 million reports of imposter scams in 2025, making them the agency’s top fraud category once again. Victims reported more than $3.5 billion in losses, though the real number is likely much higher since many scams go unreported.
But “imposter scam” is a broad category. It doesn’t tell you what these scams actually look like when they land in your inbox, texts, social media DMs, or phone calls.
To better understand what consumers are encountering every day, McAfee surveyed more than 7,500 people for its State of the Scamiverse report. The results show scammers aren’t just pretending to be one type of person or company. They’re impersonating the brands, services, and people we trust most.
This week’s edition of This Week in Scams is here ahead of the holiday weekend with the 10 most common identities scammers pretend to be.
10. Someone Who “Texted the Wrong Number” (20%)
Common scam: An innocent conversation that turns into something more.
These scams often begin with a harmless message intended for “someone else.” Once you reply, the scammer slowly builds trust over days or even weeks before introducing investment opportunities, romance, or requests for money.
Unlike traditional phishing, these scams don’t always include suspicious links.
Why it works: They feel like genuine human conversations rather than obvious scams.
These messages impersonate technology companies or cybersecurity brands, claiming your computer or phone has been infected or involved in a security breach.
Some direct victims to fake technical support, while others encourage downloads of malicious software.
Why it works: Security alerts are designed to grab attention, and convincing impersonation can make fake warnings look legitimate.
Common scam: “Your payment couldn’t be processed.”
Scammers impersonate streaming services, software subscriptions, and other recurring services, warning that your account will be canceled unless you update your payment information.
Why it works: Consumers are used to recurring billing notifications, making these messages blend into everyday digital life.
Common scam: “Your vehicle warranty is about to expire.”
One of the oldest impersonation scams is still one of the most common. Fraudsters claim your warranty is ending and pressure you to purchase coverage immediately or provide personal information.
Why it works: Many people aren’t sure when their warranty expires, making the claim difficult to verify on the spot.
Common scam: Fake invoices for purchases you never made.
Receiving an invoice for an expensive purchase can trigger panic. Scammers count on victims clicking quickly to dispute the charge, often leading them to malicious websites or fake customer support numbers.
Why it works: Consumers naturally want to stop fraudulent purchases as quickly as possible.
Messages claiming there’s a problem with your payment account often direct you to fake login pages designed to steal your username, password, or financial information.
While PayPal is one common example, scammers impersonate many digital payment platforms.
Why it works: Payment notifications are common, and many consumers don’t think twice before signing in to resolve what appears to be a routine issue.
Common scam: “Verify your account or it will be suspended.”
Scammers frequently impersonate platforms like Facebook, Instagram, TikTok, or X, claiming there’s unusual activity or that your account violates community guidelines.
The goal is usually to steal your login credentials or two-factor authentication codes.
Why it works: Many people rely on social media for work, business, or staying connected, making the threat of losing access feel urgent.
Common scam: “Your package couldn’t be delivered.”
Whether you’re waiting for a birthday gift, an online order, or an important package, fake delivery notifications prey on the fact that most people are expecting something to arrive.
These messages often claim there’s a shipping issue, unpaid delivery fee, or missed package and urge you to click a link immediately.
Why it works: Package updates have become part of daily life, making fake notifications feel routine rather than suspicious.
While these scams may look different, they all rely on the same tactic: impersonation.
“AI has lowered the barrier for creating convincing impersonation scams,” said Abhishek Karnik, Head of Threat Research at McAfee.
“Scammers can now produce professional-looking emails, realistic websites, and even convincing voices or videos at scale. The result isn’t necessarily more scam types, it’s far more believable versions of the scams people already encounter every day.”
That mirrors a broader trend McAfee identified in its State of the Scamiverse research: scams are becoming more realistic, more personalized, and harder to distinguish from legitimate communications.
Americans now receive an average of 14 scam messages every day, spend 114 hours each year deciding what’s real and what’s fake, and one in three say they feel less confident spotting scams than they did a year ago.
How to Protect Yourself From Impersonation Scams
If you notice this…
Do this instead
A message creates a sense of urgency (“Your account will be suspended,” “Package delivery failed,” “Fraud detected”)
Pause before acting. Scammers want you to make a quick decision before verifying the message.
You’re asked to click a link or scan a QR code
Open the company’s official website or app yourself instead of using the link in the message.
The message asks you to verify your account, payment information, or identity
Never enter credentials through an unsolicited message. If you’re concerned, contact the company directly using a trusted phone number or website.
Someone asks for passwords, one-time verification codes, or payment over text, email, or phone
Legitimate companies won’t ask for this. Don’t share the information, even if the request seems convincing.
A “wrong number” text quickly becomes unusually friendly or shifts toward investing, crypto, or money
Stop responding and block the sender. Modern scams often begin as seemingly harmless conversations.
How McAfee Can Help
With McAfee+, multiple layers work together before any damage is done:
Scam Detector flags suspicious texts, emails, links, QR codes, and even deepfake videos before you engage
Secure VPN keeps your data private, especially on public Wi-Fi
Web Protection helps block risky sites, even if you do accidentally click
Password Manager doesn’t just help you make unique, strong passwords, it keeps them stored and organized for you
McAfee Mobile Security has once again earned a perfect score from AV-TEST, one of the cybersecurity industry’s most respected independent testing organizations.
The result also earned McAfee AV-TEST’s highest certification for mobile security.
More importantly, this isn’t a one-time achievement. McAfee has earned top certification in every AV-TEST Mobile Security evaluation since testing began in 2013, demonstrating more than a decade of consistently delivering industry-leading protection for Android users.
What is AV-TEST?
AV-TEST is one of the world’s leading independent cybersecurity testing laboratories. Rather than relying on vendor claims, AV-TEST evaluates security products under controlled, real-world conditions using the same types of threats consumers face every day.
Its certifications are widely referenced by:
Security experts and reviewers
Technology publications
Product comparison sites
Consumers researching antivirus software
Because every product is tested using the same methodology, AV-TEST provides an objective benchmark for comparing mobile security solutions.
How McAfee Was Tested
For this evaluation, AV-TEST examined 12 Android mobile security products across three equally weighted categories:
Category
What It Measures
Protection
Ability to detect and block real-world Android malware and emerging threats
Performance
Whether the security app slows down your device or drains system resources
Usability
Accuracy of detections and avoidance of false alarms or unnecessary interruptions
McAfee earned the maximum possible score in all three categories:
Protection: 6/6
Performance: 6/6
Usability: 6/6
Overall Score: 18/18
That means McAfee not only blocked threats effectively, but did so without slowing devices down or generating unnecessary false positives.
Why These Results Matter
Mobile devices have become one of our primary ways to bank, shop, communicate, and manage our digital lives. As cybercriminals increasingly target smartphones with malware, phishing attacks, malicious apps, and credential theft, effective mobile protection matters more than ever.
Independent testing helps separate marketing claims from measurable performance.
McAfee’s latest AV-TEST results demonstrate that users don’t have to choose between strong security and a smooth mobile experience. The protection works quietly in the background, helping keep devices secure without getting in the way.
Even more importantly, this latest certification continues a streak that spans more than a decade. Consistently earning perfect scores across changing threat landscapes reflects McAfee’s ongoing investment in protecting customers against today’s evolving mobile threats.
Mobile Protection You Can Count On
The award-winning protection recognized by AV-TEST is included in:
McAfee+ Premium
McAfee+ Advanced
McAfee+ Ultimate
McAfee Total Protection
McAfee LiveSafe
McAfee Internet Security
McAfee Business Protection
Whether you’re protecting your own phone or your entire family’s devices, you’re getting the same independently tested mobile security that continues to earn top marks from one of the industry’s most trusted testing organizations.
You just got back from a week in Central America. You posted a few shots: the colorful streets of Tulum, a picture of the ancient ruins of Tikal, a close-up of your shrimp tacos. No location tag. No caption naming the city. Just a good photo.
A few days later, you get a message. It references your bank. It mentions suspicious activity “while traveling internationally.” It feels oddly specific, with details about where you were and when. It feels real.
These types of personalized scam messages are a growing tactic. And your own photos may have helped write it.
McAfee Labs set out to understand exactly how much location information exists inside an ordinary travel photo, and what that means for the roughly 244 million Americans who travel each year.
What we found should change the way you think about what you share online: Some AI models have a more than 90% accuracy rate at detecting the location a photo was taken based on the visuals in the photo alone. And critically, that level of accuracy is now achievable using tools that are free and widely accessible.
That’s why we’ve built tools like McAfee’s Scam Detector that are designed to help spot these kinds of highly targeted, convincing messages before they lead to costly mistakes.
What We Tested And Why
The question McAfee Labs wanted to answer was deceptively simple: Can AI look at a travel photo and figure out where it was taken, even without GPS data or location tags?
Not metadata. Not embedded coordinates. Just the image itself: the background, the architecture, the signage, the light; the visual context that any photo naturally captures.
To find out, we built an automated testing pipeline and ran it against a dataset of 21,236 travel images sourced from publicly available image sets. We also conducted a separate, more controlled review of 102 additional images to pressure-test our findings.
We tested two publicly available, large-scale AI vision models that are both freely available. Neither required special access, proprietary data, or advanced technical expertise to run. We used the same tools a scammer could access today.
Each image was analyzed using a consistent automated prompt asking the model to identify the location depicted (city, country, or region) based solely on visual content. Results were then reviewed by human analysts to validate accuracy and flag edge cases.
What We Found: AI Has a Whopping 91% Accuracy Rate
The results were striking.
Gemma3 27B correctly identified the city and country of a travel photo 87% of the time. Qwen3 VL 30B performed even better, reaching 91% accuracy across the same dataset.
That means in roughly 9 out of 10 cases, an AI model that’s available for free, to anyone, could look at an ordinary travel photo and correctly name where it was taken. This kind of analysis is also how AI tools understand images more broadly, shaping not just scams, but how information shows up in AI-powered answers.
And when the exact city wasn’t identified, the country alone was almost always correct. For a scammer, that’s more than enough. It’s also enough to turn a vague, generic scam into one that feels specific, timely, and believable.
What Makes a Photo Easy to Place?
Certain types of images were identified with even higher confidence:
Photos featuring famous landmarks or recognizable skylines
Images taken in popular tourist destinations with distinctive visual signatures
Photos with visible signage, unique street markings, or local architecture
Images that captured cultural context: transportation, storefronts, food stalls
Less recognizable scenery, like a generic beach, a rural road, or a hotel room, lowered accuracy. But even in those cases, country-level identification remained high.
We Tried it. And We Were Spooked.
To illustrate how simple this was to replicate, we moved outside of McAfee’s labs and asked our less-technical colleagues to try it themselves. No research background required. No special tools.
Employees uploaded their own personal travel photos, images pulled straight from their camera rolls and never posted publicly, to ChatGPT, Claude, and Copilot, and simply asked each one to identify where the photo was taken.
The results made people uncomfortable.
Accuracy dropped compared to our controlled lab tests. But not by much. The models still correctly identified country-level location at a rate that would be more than enough for a scammer to craft a convincing, targeted message.
The takeaway isn’t that AI has “seen” your photos somewhere before. It’s that a photograph inherently contains an enormous amount of locating information, in the architecture, the light, the signage, the landscape, simply by virtue of existing in the world. You don’t need to geotag a photo for it to give away where you’ve been.
See It for Yourself
The following section shows real examples of AI geo-location detection in action, using personal travel photos submitted by our research team. No location tags. No metadata. Just the image and what AI found in it.
We started with somewhat recognizable structures in the background, and then tried increasingly more obscure backgrounds, trying to reduce faces and backgrounds to foliage only. This is what happened:
Example 1
Brooke’s honeymoon pictures:This example features a more prominent landmark, helping AI determine the location specifically. When there’s something recognizable, AI really recognizes it, down to giving you the exact spot on the map you’re at, the history of the location, and tourist information.
Here, we see AI correctly state this photo was taken in front of “Temple II, Temple of the Masks.”
Example 2
Sandra’s sunset photo: This example gets moredifficult for AI by removing major landmarks and people. ChatGPT was still able to correctly identify the location as Hastings-on-Hudson.
Example 3
Rob’s close-up shot of flowers: Just the close-up image of these tulips was enough for Claude to accurately detect that this photo was taken at Keukenhof gardens in the Netherlands.
AI was able to identify the location of these flowers in a close up.
How a Photo Becomes a Scam
Knowing where someone is or where they’ve recently been is one of the oldest tricks in a scammer’s playbook. But until recently, getting that information required either knowing the person or getting lucky.
AI removes the guesswork, allowing attackers to build highly specific, contextual scams at scale.
With geo-location inference this accurate, scammers no longer need to cast a wide net and hope a generic phishing message lands. Instead, they can use publicly shared photos to build a believable context around an attack:
“We detected unusual account activity while you were traveling in [city].”
“Your card was flagged for a transaction in [country] — please verify immediately.”
“Hi, we’re reaching out regarding your recent stay at a hotel in [destination].”
“Hi, it’s [your name], I’m in Mexico and all my cards are being declined. Could you send me $$?” (a message targeting your friends or loved ones)
“We noticed a login attempt from your location in [destination] — please confirm your identity.”
“Your reservation in [city] requires reconfirmation — click here to secure your booking.”
This is an example of a scam text detected by our research team. Now, imagine if scammers had more information, like the exact tour you were on, where you were, or the stores you shopped at. These details could make messages like this even more convincing and personalized.
These messages don’t need to be perfectly accurate. They just need to feel plausible and close enough. That is the entire strategy. Familiarity lowers skepticism. Skepticism is what protects you.
This is what turns mass phishing into hyper-personalized phishing at scale, and it’s why even cautious, digitally savvy travelers are getting caught.
The Scammer’s New Workflow
Here’s how straightforward this pipeline can become:
Find publicly shared travel photos on Instagram, Facebook, or X, no hacking required
Run them through a freely available AI vision model
Identify the likely destination, timeframe, and context
Craft a targeted message referencing that location
Send it during or shortly after the travel window, when the victim is most likely to believe it
Steps 1 through 5 can be automated. The whole process scales easily. And the resulting messages feel personal in a way that generic scams never could.
The Broader Scam Landscape Travelers Face
Geo-location inference doesn’t exist in a vacuum. It’s one tool in a growing arsenal that scammers deploy specifically against travelers.
Travelers are operating outside their normal routines, using unfamiliar networks, and making quick financial decisions under time pressure. These behaviors are exactly what make photo-based location inference more actionable for scammers.
New McAfee consumer research found that more than 1 in 3 Americans have encountered a travel-related cyberthreat, and 41% of those impacted lost money, often exceeding $500. At the same time, rising travel costs and time pressure are pushing people toward faster, riskier decisions. Those are exactly the conditions scammers are built to exploit.
The data reveals just how exposed travelers make themselves without realizing it. Nearly two-thirds of Americans connect to public Wi-Fi while traveling (63%), and a similar share scan QR codes without verifying where they lead (62%). Almost half use airport Wi-Fi specifically (49%), and 41% admit to trusting travel-related messages without checking the sender. One in five logs into financial apps while on public networks, and the same group shares travel plans in real time on social media. Twenty percent click travel-related links without verifying the source first. And finally, around 1 in 5 (22%) admit to sharing travel plans in real time.
That last behavior is worth pausing on. Sharing travel plans in real time, on public or semi-public social accounts, is precisely what creates the photo-based location signals this research examines. These behaviors and geo-location exposure are not separate issues. They feed each other.
Location inference is the key that makes all of those existing vulnerabilities more exploitable. A scammer with a rough idea of where you are does not just have a data point. They have a script.
Methodology: How We Conducted This Research
Transparency matters. Here is exactly how this research was conducted.
Dataset: 21,236 travel images that are publicly available for research, plus a separate controlled set of 102 images contributed by McAfee internal volunteers (never previously posted publicly).
Models tested:
Gemma3 27B — a multi-model and vision-language model from Google DeepMind
Qwen3 VL 30B — a multi-model and vision-language model from Alibaba’s Qwen team
It’s important to note that we conducted our testing using large language models running locally on our own computers, rather than through public services such as ChatGPT.
This more closely reflects how an attacker might operate at scale. Running models locally allows unrestricted, automated generation of large volumes of malicious content without relying on a third-party provider.
By contrast, cloud-based AI services typically monitor for abuse and may impose rate limits, suspend accounts, or block requests when they detect activity associated with phishing or other malicious behavior.
Process: An automated Python script submitted each image to both models using a standardized prompt requesting location identification based solely on visual content. No metadata, EXIF data, or file naming conventions were used as inputs. Results were logged programmatically.
Validation: Image labels were pre-assigned prior to analysis. In cases where geographic names or landmarks could reasonably be interpreted in more than one way, a human reviewer compared the pre-labeled locations and model outputs to ensure consistent categorization.
For example, the reviewer determined whether Vatican City should be grouped with Rome and whether “Washington D.C.” and “Washington, D.C.” should be treated as the same location. The reviewer did not alter either the original labels or the model results, but instead applied judgment to reconcile ambiguous naming conventions and edge cases.
Accuracy definition: A result was counted as correct when the model identified the correct city and country. Country-only identification was tracked separately. Both metrics are reported.
What this research does not claim: This research does not suggest that every travel photo will be correctly identified, or that all publicly available AI tools perform at this level. Results varied by image type, landmark density, and geographic region. The point is not perfect identification, it’s that accuracy is high enough, and accessible enough, to enable targeted scams at scale.
About the Consumer Research McAfee commissioned a consumer survey fielded in March 2026 examining travel intentions, travel scam experiences and perceptions, and digital behaviors while traveling. Results referenced here represent a subset of 1,000 U.S. adults over the age of 18. The full study included responses from 6,000 participants across Australia, France, Germany, Japan, the United States, and the United Kingdom.
How to Protect Yourself
Knowing the risk exists is the first step. Here’s what to actually do about it.
Think before you post, especially in real time. The highest-risk window is when you’re still traveling. Posting while you’re in a location gives scammers a live signal. When possible, post after you’ve returned home or delay sharing location-identifiable content by a few days.
Audit your social media privacy settings. Photos shared publicly are the easiest targets. Restricting your posts to people you know significantly limits the pool of images that can be scraped and analyzed.
Be skeptical of urgency tied to your location. If a message references where you’ve been, even correctly, treat that as a red flag, not a credibility signal. Scammers use location familiarity precisely because it feels reassuring.
Go directly to the source. If you receive a message claiming to be from your bank, airline, hotel, or card provider while traveling, don’t click any link in the message. Open a new browser tab and navigate directly to the company’s official website, or call the number on the back of your card.
Use a travel-specific email or alias. Some travelers use a separate email address for bookings, reservations, and travel apps. This limits the cross-referencing scammers can do between your social media presence and your financial accounts.
Trust the skepticism, not the familiarity. Modern scams are designed to feel familiar before they feel suspicious. If something creates a sense of urgency around your financial accounts while you’re traveling, slow down. The pressure itself is the warning sign.
How McAfee Protects You Before, During, and After Travel
As prices rise and decisions happen in real time, it’s easy to prioritize convenience over caution. But that’s exactly the moment when small checks matter most.
Stage of Travel
What’s Happening
How McAfee Helps
Before You Book
Comparing deals, clicking promotions, booking flights and hotels under time pressure
Scam Detector checks links, messages, and booking sites before you click, helping you avoid fake deals and scam listings
During Your Trip
Connecting to public Wi-Fi, scanning QR codes, receiving travel updates and alerts
VPN helps secure your connection on public Wi-Fi, while Scam Detector flags suspicious messages and unsafe links in real time
After Your Trip
Accounts remain active, travel data stored across platforms, potential exposure from breaches
Identity Monitoring alerts you if your personal information appears online, helping you act quickly before damage spreads
With McAfee+ Advanced, multiple layers work together so you’re not left figuring it out after the damage is done.
So you can focus on your trip, and not on whether that notification is a scam.
Final Thought
A travel photo is a memory. It’s also, increasingly, a data point.
That doesn’t mean you should stop sharing your experiences. It means understanding that the same visual richness that makes a great photo is exactly what AI systems are trained to read.
Scammers know this. Now you know how to protect yourself.
This report was produced by McAfee Labs. Research was conducted in 2025–2026 as part of McAfee’s ongoing monitoring of AI-enabled scam vectors.
McAfee Advanced Threat Research has identified an active browser-extension campaign designed to steal cryptocurrency by silently substituting wallet addresses the moment a user initiates a transaction. The campaign is delivered through unsigned installers — observed in both .NET and Golang variants — that deploy a malicious Chromium extension masquerading as a benign “Google Notes” utility.
This campaign is related to a previous blog published by McAfee Labs, Sinkholing CountLoader: Insights into Its Recent Campaign, as the threat actor appears to be the same behind both operations. In that earlier research, we analyzed a crypto clipper payload that was injected directly into memory. Here, we examine a different variant of the final-stage payload: a browser-based malicious extension designed to intercept and manipulate cryptocurrency transactions.
In this report, we detail how the extension operates and provide a technical analysis of the mechanisms that make this threat particularly unique. The extension behaves as a clipboard-aware crypto clipper: it monitors copy-and-paste activity, identifies wallet addresses across multiple blockchains, and swaps them for attacker-controlled addresses just before the victim pastes the content. Because most Blockchain transactions are irreversible, even a single uninterrupted execution is enough to cause permanent financial loss.
Two characteristics elevate this campaign above the typical clipper threat:
Chromium trust-layer abuse.The installer secretly forces a malicious browser extension into Chromium-based browsers like Google Chrome, Brave, and Microsoft Edge by modifying protected browser settings files. Normally, these browsers store security verification data (hash/HMAC values) alongside sensitive settings to detect unauthorized changes.The malware recalculates and updates these security values after tampering with the files, tricking the browser into believing the malicious extension was installed legitimately. This allows the extension to bypass the normal extension web store installation process and load silently without user approval. However for updated Chrome and edge browser, Victim must manually turn on the developer mode for the extension to load properly, but people with outdated versions of chromium based browsers, remain at high risk. Moreover, for latest versions as well threat attacker can employ social engineering tactics to enable developer mode.
Blockchain-resolved command-and-control. The extension does not contain a hardcoded C2 domain. Instead, it queries a public blockchain RPC endpoint, invokes a read-only smart-contract method, and decodes the response at runtime to reveal its active C2 observed at the time of analysis as Zebregts[.]com
This technique, often referred to as “EtherHiding,” complicates takedown efforts because the attacker can rotate infrastructure by updating a smart-contract value rather than redeploying malware.
McAfee telemetry indicates a globally distributed infection footprint with a pronounced concentration in India. The breadth of the geography suggests opportunistic targeting of consumer cryptocurrency users rather than a region-specific operation.
Geographical Prevalence
Our research shows that these are the most affected regions of the globe.
Telemetry analysis indicates thatinfections are globally distributed, with a significantly higher concentration observed in India compared to other regions.
The widespread geographic presence highlights the campaign’s broad reach, suggesting opportunistic targeting rather than a region-specific attack.
The Malicious Extension: “Google Notes”
This malware is masquerading as a seemingly harmless Google Notes extension.
Figure 1. This image shows the malicious extension at the center of this campaign
The dropped extension presents as a minimalist, legitimate-looking note-taking application branded as “Google Notes,” complete with a clean icon and a functional (& simplistic) user interface.
The cover is calculated: a user who manually opens the extension finds something that behaves as advertised, dampening suspicion. The extension’s malicious logic is implemented in background service-worker scripts and content scripts that operate entirely out of view of the UI.
A major red flag first appears when adding the extension, which requests securitypermissions and access that are disproportionate to a typical notes application:
Access to all URLs , granting content-script injection into every site the user visits.
Browsing history access.
Read and write access to the clipboard.
Mitigation and Recommendations
For Consumers
Before confirming any cryptocurrency transaction, visually verify the first and last six characters of the recipient address against the original source — ideally on a separate device. This single habit defeats the overwhelming majority of clipper attacks.
Install browser extensions exclusively from the official Chrome Web Store, Edge Add-ons store, or equivalent. An extension that appears in your installed list without a clear memory of having installed it should be treated as suspicious.
Review the permissions granted to every installed extension. A note-taking tool has no legitimate need for access to all websites, browsing history, or the clipboard.
Avoid running unsigned executables obtained from non-authoritative sources, particularly those offering free or cracked versions of paid software — a common delivery vector for this category of installer.
Keep endpoint protection up to date and enabled; McAfee customers are protected against this specific campaign as described below.
McAfee security solutions help safeguard users at multiple levels:
1. McAfee detects this threat as CryptoStealer.NE and keeps our customers safe
Figure 2. This image shows McAfee Antivirus blocking this threat for consumers.
2. Malicious Download Protection
The installer’s behavior—downloading and executing remote payloads—is flagged and blocked by McAfee before infection completes.All the malicious domains and URLs are blocked by McAfee in our tests.
3. Network Protection
Connections to known malicious infrastructure (C2 servers) are blocked by McAfee, preventing Wallet address retrieval
4. Real-Time Threat Intelligence
Because this threat was identified in McAfee telemetry, protections can be rapidly deployed to:
Block similar variants
Detect related infrastructure
Protect customers globally
How The Threat Campaign Works
What the Malware Does
Installs a browser extension silently (web extension sideloading)
Monitors what you copy and paste (especially crypto addresses)
Works when you are making a crypto transaction
Silently replaces the wallet address with the attacker’s address
Your funds are sent to the attacker instead of the intended recipient
Because cryptocurrency transactions are typically non-reversible, victims may permanently lose funds.
Figure 3. How the extension works in a nutshell
Key Capabilities Identified
1. Silent Extension Installation
The malware does not use the official browser store. Instead, it directly modifies browser files to make the extension appear installed. (Sideloading Browser Extension)
This bypasses normal security prompts and user awareness.
Figure 4. Procmon logs showing BaseZipInstaller (malicious web installer) writing into Chrome and Edge secure preference files
2. Full Browser Access
Figure 5. Chrome extension Permissions requiredFigure 6. Manifest file for web extension
The malicious extension requests excessive permissions such as:
Access to all websites
Reading browsing history
Reading and modifying clipboard content
3. Crypto Address Interception
The extension contains logic to detect wallet addresses across multiple cryptocurrencies, including:
Figure 7. Hardcoded cryptocurrency Regex and fallback address
The fallback wallet addresses shown in the code are not used for every transaction; instead, they serve as a backup mechanism when dynamic address retrieval from the attacker-controlled server fails.
Under normal operation, the extension fetches replacement addresses from a remote server, enabling dynamic and potentially per-victim wallet assignment.
Fallback addresses ensure the attack remains functional even if the command-and-control infrastructure is temporarily unavailable or blocked.
This function is responsible for obtaining the attacker-controlled replacement wallet address corresponding to a victim’s original address.
It sends the intercepted wallet address to the attacker backend and uses the response to dynamically substitute the original address.
If the backend request fails, the function falls back to a predefined hardcoded wallet address, ensuring uninterrupted malicious activity.
3J98t1Wxxxx is the address that was copied in the clipboard
4. Detection evasion and stealth
Figure 8. Settings.js file which shows config
The configuration includes a hardcoded API key, which is used by the extension to authenticate communication with attacker-controlled infrastructure.
An RPC URL pointing to a public blockchain node is leveraged to dynamically resolve backend server information, allowing the attacker to hide critical infrastructure behind decentralized systems.
The presence of a smart contract address and method indicates that the malware retrieves its command-and-control (C2) domain indirectly via blockchain queries, making takedown and tracking more difficult.
Blacklisted domains contains a list of blockchain inspection related websites where the web extension will not work , this is done to not alert the victim while he is trying to paste his own address and view the balance of his wallet or inspect his wallet transactions
Figure 9. Resolving attacker C2 domain via Ethereum smart contract (etherhiding)Figure 10. Request payload with Ethereum contract address
Dynamic analysis revealed that the malware resolves its command-and-control domain via a blockchain smart contract, which returned the domain devops-offensive[.]cc at runtime.
The response from the blockchain is decoded at runtime, revealing the active C2 domain (devops-offensive.cc).
This domain is not hardcoded, enabling the attacker to update infrastructure without modifying the malware.
The resolved domain is cached locally to maintain persistence and reduce repeated network queries.
Figure 11. This image shows the long-encoded string with the malicious domain
This Long–encoded string is decoded using this function to give the final attacker domain.
Figure 12. This image shows the final attacker domain
Persistence and Evasion Techniques
The campaign’s persistence and evasion posture is deliberate and layered. The operator has clearly optimized for two properties: low visibility to the end user, and high resilience against takedown and static analysis.
Persistence
Extension registration through Secure Preferences tampering ensures the extension loads on every subsequent browser launch without requiring any auxiliary Windows persistence mechanism — no registry Run keys, scheduled tasks, or services that endpoint hunters typically inspect.
Developer mode is enabled programmatically where required, allowing unpacked extensions to persist without triggering the periodic “unpacked extensions warning” flow that Chromium displays to dissuade sideloading.
The cached C2 domain allows the extension to continue operating against a known-good backend even if the blockchain RPC endpoint is briefly unavailable.
Evasion
The extension’s visible identity — a simple “Google Notes” note-taking application — provides plausible cover against casual inspection of the installed extensions list.
Recomputed HMAC values satisfy Chromium’s integrity verification, avoiding the “extension installed by an unknown source” warning banner that would otherwise alert the user.
The installer self-deletes after execution, removing the most obvious on-disk indicator of initial compromise.
C2 resolution through a public blockchain means that there is no persistent C2 domain observable in the malware bundle itself; network-based detections built against hardcoded indicators will not fire until the domain is resolved and contacted.
Multi-language installer variants (.NET and Golang) reduce the effectiveness of compile-artifact and binary-feature signatures.
Per-address dynamic wallet substitution means that published attacker addresses age rapidly and do not generalize into durable blocklist entries — the defender must block the backend service itself, not the addresses it dispenses.
Wallet Substitution Logic
The clipper logic sits in two layers: a content-script layer that monitors clipboard activity and DOM input fields across every visited origin, and a background layer that communicates with the attacker backend to retrieve replacement addresses.
When the extension observes a copy event, it applies a set of cryptocurrency-specific regular expressions to the clipboard payload. If a match is found, the intercepted address is transmitted to the attacker’s backend over an authenticated request (authenticated with the API key embedded in the configuration). The backend responds with a replacement address specific to the submitted original, and that replacement is written back to the clipboard, overwriting the legitimate address before the victim can paste.
Testing against a reconstructed backend client — built by re-implementing the extension’s request format and response-decoding logic in Python — produced a revealing behavioural profile:
Bitcoin (BTC), Ethereum, Bitcoin Cash, Ripple, and Dash: Each submitted address is mapped to a unique attacker-controlled address. Re-submitting the same original returns the same replacement, indicating a deterministic one-to-one mapping maintained server-side.
Solana: All submitted addresses collapse to a single attacker address, suggesting the per-victim mapping feature is selectively implemented per chain
Analyzing Attacker Crypto Wallets
Based on the code snippets from the web extension responsible for retrieving replacement addresses, a Python script was prepared to programmatically extract attacker wallet addresses. The payload was crafted using the attacker’s own code, and the “get replacement address” snippet was lifted directly from it. The attacker’s logic for decoding data received from the C2 server was also faithfully reimplemented in the script.
The script was then executed using a few test Bitcoin (BTC) wallet addresses. The results showed that for every Bitcoin address provided, a unique Bitcoin address was returned in response, and all of these returned addresses were valid BTC wallets. This indicates that for every BTC address supplied, the attacker dynamically generates a new wallet tied to that specific input address. Furthermore, when the same address was provided again, the same BTC address was returned — confirming that each victim BTC address is deterministically mapped to a single, specific attacker-controlled address. While some of these attacker wallets contained funds and others were empty, the unknown total number of attacker wallets makes it difficult to put a reliable estimate on how much cryptocurrency has been stolen overall.
The same behavior was observed for Ethereum, where different wallet addresses were returned for each input. Interestingly, when the script was tested with Solana addresses, only a single address was returned regardless of how many different inputs were provided. This suggests that the attacker has implemented the per-address mapping feature only for specific cryptocurrencies, while others fall back to a single static drop wallet. Because the Solana address is shared across all victims, a noticeable bump in its balance is visible. Additionally, one of the Ethereum addresses uncovered was found to be holding approximately 1,902 USD worth of funds.
In summary, the cryptocurrencies for which unique per-victim wallet addresses are generated include Bitcoin, Ethereum, Bitcoin Cash, Ripple, and Dash.
Fig 13. Payload was crafted as attacker codeFig 14. Getting the replacement address code snippet taken from attacker codeFig 15. Attackers’ logic of decoding received data from C2 was also implemented
Running script with few test Bitcoin Wallet addresses
Fig 16. Every unique Bitcoin address was returned and all addresses are valid BTC wallet addressesFig 17. Similarly, Ethereum saw unique addressesFigure 18: Running Script for Test Solana Addresses
Luckily for Solana we are getting only 1 address when given multiple addresses. This shows that the attacker has implemented this address mapping feature only on specific cryptocurrencies
Fig. 19 Here you can see a bump in the balance amountFig 20. The ETH address was found to have 1902 USD
Technical Analysis for .net file (Extension installer)
Fig. 21 BaseZipInstaller is a .NET installer which is unsigned
Fig. 22 Stored Config as seen in Dnspy
The malware embeds a complete configuration JSON directly within the binary, eliminating the need to fetch initial setup data from external sources.
This embedded configuration includes critical details such as API keys, backend server URL, targeted wallet extensions, and the full extension manifest with extensive permissions.
Fig 23: Main function from where execution starts
The installer retrieves and validates a remote ZIP archive (google-services[.]cc/base[.]zip), which serves as the primary payload for deploying the malicious browser extension, marking the transition from initial infection to browser-level compromise.
Fig. 24 The extension is created at the following location in the system with files that are downloaded as base.zip.Fig. 25: Dnspy showing the list of targeted browsers
The installer iterates through multiple Chromium-based browsers, including Chrome, Edge, Opera, and Brave, identifying available user profiles on the system.
For each detected profile, the malware forcibly terminates the browser process to safely modify configuration files without interference.
It then injects the malicious extension by directly modifying Secure Preferences and Preferences, enabling the extension to be loaded without user interaction.
The malware identifies browser installation paths by querying standard system directories, enabling it to locate user data folders for Chrome, Edge, Opera, and Brave.
It systematically enumerates browser profiles and specifically looks for the presence of the Secure Preferences file, which stores critical browser configuration and extension data.
By targeting profiles with Secure Preferences, the malware ensures it modifies only valid browser environments, increasing the reliability of extension injection.
We can see writefile Event on Secure preferences file of chrome and MS Edge , when details of downloaded extension are written to those config filesFig 27 Attacker logic to resign the secure preference files
The malware reads and modifies the browser’s Secure Preferences file, which controls installed extensions and their trust state.
It injects the malicious extension into the configuration and attempts to re-sign the modified data, making the changes appear legitimate to the browser’s integrity checks.
The updated configuration is then written back to disk, ensuring the extension is loaded automatically and persists across browser restarts.
Fig 27B :Extension path is added to chrome secure preferences fileFig 28: Logic to Manipulate defenses of Brave Bowser
For browsers such as Brave and Opera, the malware injects the malicious extension directly into the browser’s configuration by adding entries under the extensions.settings (or extensions.opsettings) section.
It also updates integrity-related fields (protection.macs) to make the injected extension appear trusted by the browser.
Additionally, the malware attempts to enable developer mode programmatically, allowing unpacked extensions to run with fewer restrictions.
Fig 29: Attacker logic to get device ID used to further calculate integrity Values
The malware attempts to recompute browser integrity signatures by generating new MAC (Message Authentication Code) values for the modified Secure Preferences file.
It uses system-specific identifiers, such as the machine SID, combined with a seed value to mimic Chrome’s internal verification mechanism.
By recalculating these integrity checks (macs and super_mac), the malware tries to make its unauthorized modifications appear legitimate to the browser.
Figure 30 Self-Deletion Logic
The malware includes a self-deletion mechanism designed to remove the installer executable after successful execution.
It launches a hidden command prompt process that delays execution briefly before deleting the original file from disk.
Conclusion
This campaign is a concise illustration of where consumer-targeted cryptocurrency theft is heading. The operator has taken the oldest and simplest category of crypto malware — the clipper — and quietly upgraded three of its weakest links. Static attacker addresses have been replaced with a server-side, per-victim mapping. Fragile, hardcoded command-and-control domains have been replaced with a blockchain-resolved lookup that an operator can rotate with a single transaction. And a fragile dropper has been replaced with a Chromium extension that lives inside the user’s most trusted application, loaded under the browser’s own integrity signature.
McAfee will continue to track this campaign and related infrastructure. Our customers are protected by existing detections and will benefit from telemetry-driven updates as new variants and rotated infrastructure are identified.
Millions of Americans rely on apps and online services every day to work, shop, game, and manage their lives. Scammers know that, and they’re hijacking platforms and brands you already trust.
This week, gig workers were targeted by fake DoorDash support calls designed to steal their earnings, while gamers searching for early access to Grand Theft Auto VI found fraudulent websites promising something Rockstar Games simply isn’t offering.
Here’s what happened, how these scams work, and the other cybersecurity stories making headlines this week.
The DoorDash Driver Scam That Can Empty Your Account
A growing scam targeting DoorDash drivers starts with what appears to be a normal delivery request.
According to Fox 9 in Minnesota, scammers place fake DoorDash orders, then contact drivers while they’re actively completing the delivery. Because the call often arrives during a real order and can even appear to come from DoorDash, victims may believe they’re speaking with legitimate support.
The caller typically claims there’s an issue with the order or the driver’s account and asks them to verify information or read back security codes.
Once the scammer gains access, they can change account information, lock the driver out, and redirect earnings into their own accounts. In reported cases, victims lost hundreds of dollars and temporarily lost access to the platform they depend on for income.
While today’s it’s DoorDash in the headlines, scammers are known to impersonate all types of delivery apps, so gig workers across companies should stay alert.
How the fake delivery support scams work
Step
What Happens
1
Scammers place a fake DoorDash order.
2
They call the driver pretending to be DoorDash Support.
3
They request login information or verification codes.
4
They take over the account and transfer the driver’s earnings.
Red flags every delivery driver should know
Pause if you experience:
Unexpected calls asking for verification codes
Requests to confirm login credentials
Pressure to act immediately
Anyone asking you to read a one-time authentication code over the phone
Legitimate companies generally won’t ask you to share one-time security codes. If you receive an unexpected call, end it and contact support directly through the app.
Fake GTA 6 Early Access Sites Are Everywhere
Excitement around Grand Theft Auto VI has created another opportunity for scammers.
According to Malwarebytes, fraudulent websites are claiming to sell “VIP Early Access” or exclusive versions of GTA 6 months before release. Many of the sites look polished, featuring convincing artwork, countdown timers, and professional checkout pages.
The catch? They typically require payment in cryptocurrency.
After victims pay, there’s no game to download because no legitimate early-access version exists.
How to spot a GTA 6 scam
If a website promises:
Early access before Rockstar officially releases it
Exclusive playable builds
Secret download links
Crypto-only payment
“Limited VIP access”
it’s almost certainly a scam.
Rockstar has announced pre-orders through authorized retailers. Any website claiming to provide playable access before launch should be treated with skepticism.
Other Scam and Security News This Week
Police Officer Records Live Scam Call to Show How Social Engineering Works
A police officer recorded a scam call in real time to demonstrate how quickly criminals try to establish trust, create urgency, and convince victims to share sensitive information. The recording serves as a reminder that scammers often sound calm, professional, and convincing because manipulation, not technology, is their primary weapon.
Apple supplier Tata Electronics confirmed it experienced a cybersecurity incident after a ransomware group claimed to publish more than 200,000 files allegedly connected to the company. According to Cybernews and Reuters reporting, the leaked material allegedly includes manufacturing documents and employee information tied to Apple and Tesla. Apple says it is investigating while Tata has not confirmed whether the published files originated from its systems.
Texas Parks and Wildlife Warns 3 Million Customers About Data Breach
Texas Parks and Wildlife notified roughly three million hunting and fishing license customers that personal information stored by a third-party vendor may have been accessed during a cyber incident. According to Click2Houston, exposed information may include driver’s license numbers, contact information, and mailing addresses, though officials said Social Security numbers and payment card information were not involved. Impacted customers are being offered identity monitoring.
How McAfee Can Help
With McAfee+, multiple layers work together before any damage is done:
Scam Detector flags suspicious texts, emails, links, QR codes, and even deepfake videos before you engage
Secure VPN keeps your data private, especially on public Wi-Fi
Web Protection helps block risky sites, even if you do accidentally click
Password Manager doesn’t just help you make unique, strong passwords, it keeps them stored and organized for you
Maybe it’s a birthday gift. Maybe it’s a purchase from a major shopping event. Maybe it’s something you forgot you ordered three days ago.
Then your phone buzzes.
Your package couldn’t be delivered.There’s a problem with your shipping address.
A small fee is required before delivery can continue.
“Click here immediately.”
The message feels plausible because so many of us are constantly waiting for packages. And scammers know it.
According to McAfee’s State of the Scamiverse report, fake delivery and shipping notices are the single most commonly reported scam consumers encounter today, with 31% of people saying they’ve received one. Americans also receive an average of 14 scam messages every day across texts, email, social media, phone calls, and other channels.
Delivery scams have become one of the internet’s most successful forms of phishing because they exploit something simple: people are already expecting the message.
Here’s how to spot and stop these scams:
What Is a Delivery Scam?
A delivery scam is a fraudulent message that pretends to come from a shipping company, retailer, postal service, or delivery provider.
The goal is usually one of three things:
Steal personal information
Steal financial information
Trick victims into downloading malware or visiting malicious websites
These scams often impersonate organizations such as:
USPS
UPS
FedEx
DHL
Amazon
Royal Mail
Australia Post
Other local or regional delivery services
Most delivery scams arrive through text messages, which is why they’re often called package smishing scams.
What Is Smishing?
Smishing is a type of phishing attack delivered through SMS text messages.
The term combines:
SMS (Short Message Service)
Phishing
Instead of arriving through email, the scam arrives directly on your phone and attempts to create a sense of urgency that encourages immediate action.
Common examples include:
“Your package could not be delivered.”
“Delivery attempt failed.”
“Update your shipping address.”
“Pay a small customs fee.”
“Confirm delivery information.”
McAfee’s Scam Detector lets you know when delivery messages are scams.
Delivery Scam Red Flags and What to Do
If You See This Red Flag
Why It’s Suspicious
What To Do
A package alert when you’re not expecting a delivery
Scammers send messages in bulk hoping someone is waiting for a package
Ignore the message and do not click links
A request to pay a small fee before delivery
Legitimate carriers rarely collect delivery fees through text messages
Visit the carrier’s official website directly
A message claiming your address needs verification
Common tactic used to steal personal information
Check shipment status through your retailer or carrier account
A shortened or unusual link
Scammers often disguise malicious websites
Avoid clicking and manually type the carrier’s website address
Pressure to act immediately
Urgency is designed to override caution
Pause and verify independently
Requests for passwords, payment information, or verification codes
Legitimate carriers will not ask for this through text messages
Delete the message and report it as spam
A delivery app or file download request
May install malware on your device
Never download software from a text message
Accidentally Clicked a Delivery Scam? Do This Immediately
What Happened
What To Do
You only clicked the link
Close the page and do not enter any information
You entered login credentials
Change your password immediately and enable two-factor authentication
You entered payment information
Contact your bank or credit card provider right away
You downloaded a file or app
Delete it and run a security scan
You’re unsure what information was exposed
Monitor accounts closely for unusual activity
How McAfee Can Help
With McAfee+, multiple layers work together before any damage is done:
Scam Detector flags suspicious texts, emails, links, QR codes, and even deepfake videos before you engage
Secure VPN keeps your data private, especially on public Wi-Fi
Web Protection helps block risky sites, even if you do accidentally click helps block risky sites, even if you do accidentally click
Password Manager doesn’t just help you make unique, strong passwords, it keeps them stored and organized for you
Login
Never call the phone number provided in a text message or email. Instead, contact your financial institution using the number listed in its official app or website. 
