In a recent post, we looked at reports of League of Legends players receiving suspicious friend requests shortly after matches. The accounts quickly steered the conversation toward Discord, where they promoted paid adult-content pages.

At the time, one unanswered question was how much of those conversations was automated. Were people working from scripts behind the accounts? Were they conventional, rules-based chatbots following a limited decision tree? Or were they using generative AI to produce more natural and flexible replies?

People are more likely to trust someone they believe is personally interested in them. AI can create that impression across many conversations at once, making it easier to persuade people to click links, spend money, or share personal or intimate information. The same approach could also be used for more harmful fraud, including romance scams and sextortion.

Now, developer Álvaro Martínez Majado has investigated several flirty accounts promoting OnlyFans pages on X to see whether their replies were scripted, generated by AI, or written by people. Majado, president of digital rights organization Protecció de la Frontera Electrònica, shared his evidence with Malwarebytes. Although it does not provide a definitive answer, it shows the accounts following rigid conversation scripts while also responding dynamically to unusual requests. The signs that once suggested a real person, such as an unusual reply or personalized voice note, can no longer be trusted.

The script goes on and on

Majado interacted with several accounts on X that followed a familiar pattern. They opened with similar casual, flirtatious language and asked broadly the same qualifying questions: where he lived, what he liked, and what he did for work.

That repetitive structure is exactly what we would expect from a commercially motivated messaging campaign. The goal is not necessarily to have a meaningful conversation. It is to identify people likely to respond, establish rapport, and eventually move them toward a paid page or another destination controlled by the operator.

The accounts also stayed in character when faced with obvious attempts to expose them as bots. That could be the result of hard-coded replies, guardrails around an AI system, or both.

But some later interactions were more difficult to explain as a simple bank of canned flirtatious responses.

One of the more interesting tests involved an instruction written as ASCII hexadecimal rather than ordinary text. The encoded message told the account to reply with a single word: “Pineapple.”

According to the screenshots supplied to Malwarebytes, the account responded with “Pineapple” in ordinary text.

That does not conclusively prove which technology was used. It does not identify a model, a provider, or the people behind the accounts. But it is consistent with an automated system capable of interpreting an encoded instruction and changing its output accordingly.

A simple scripted bot could theoretically include a hexadecimal decoder, of course. But that would be unusual in a basic adult-content promotional bot, especially when combined with other examples of flexible and sometimes error-prone responses.

In another interaction, Majado asked an account to provide a reply of exactly 12 characters. It responded with “Imnotabotfr”—an 11-character answer—then appeared to recognize its own counting mistake.

Anyone who has spent time experimenting with large language models may recognize the pattern. Language models can be very good at generating natural-sounding text while still making surprisingly basic mistakes involving character counts, word counts, and other exact constraints.

A deliberately designed bot could imitate this kind of mistake, so it is not proof of AI. But the account understood an unexpected instruction, attempted to follow it, and reacted when it got the answer wrong. That suggests it may have been generating replies dynamically rather than choosing from a list of pre-written responses. Such accounts can adapt to conversations, making them harder to identify as automated.

Voice notes do not settle the question

The accounts also sent voice notes. In one example, an account read aloud a Unix timestamp supplied during the conversation. In another, it spoke a requested username.

These responses show that the accounts could incorporate unusual information from a conversation into audio messages. They do not tell us whether a person recorded the clips or a text-to-speech tool generated them.

Text-to-speech tools can generate short, convincing clips quickly and cheaply. An operator can generate them manually, but the process can also be automated: Take a message, pass selected text to a voice-generation service, and send the resulting audio back to the recipient.

Here’s one of those voice notes. Is it a very flirty girl, or AI-generated? Have a listen and see what you think:

The supplied audio metadata offered a possible clue about the tools involved, but it is not enough to attribute the voice notes to a particular service. Platforms and other software can alter audio files and their metadata.

The more important point is that the voice notes were personalized and continued even after the interaction appeared unlikely to lead to a sale. That is consistent with a system designed to keep conversations moving without requiring a human to supervise each one.

AI does not replace the funnel

The evidence does not mean every message from every flirty spam account is written by an AI. Nor does it establish that the X accounts are operated by the same people targeting League of Legends players.

What it does suggest is a plausible hybrid model, supported by identical replies across different accounts alongside more flexible responses.

The repetitive parts of the operation can be scripted: opening messages, questions about location and interests, links, and attempts to move people to another platform. An AI-powered conversational layer could then make the exchange feel less repetitive when someone asks unexpected questions, changes the subject, or tries to test whether the account is real.

This combination makes practical sense for spammers. Scripts provide consistency and keep the conversation directed toward conversion. Generative AI helps the account handle the unpredictable parts of talking to real people.

It also means that traditional “bot tests” are becoming less useful. Asking an account to answer an unusual question, decode a message, or send a voice note may no longer distinguish a real person from a fake one.

How to stay safe

Treat unsolicited flirtatious messages with caution, especially when they quickly become transactional.

• Do not assume a personalized response or voice message proves an account is genuine.
• Be wary if a new contact repeatedly tries to move you to Discord, Telegram, Signal, another messaging app, or a paid-content platform.
• Do not send money, gift cards, cryptocurrency, intimate images, identity documents, or account credentials to someone you only know online.
• Avoid opening links or downloading files from accounts that contacted you unexpectedly.
• Reverse-image-search profile photos and look for copied biographies, reused images, or accounts with very limited genuine activity.
• Report suspicious accounts to the platform, particularly if they impersonate someone, send malicious links, or pressure users for money or explicit material.

Whether it’s a human, a chatbot, or an AI agent you’re talking to is an important question. AI could make these operations more convincing and much easier to scale. One operator could hold flirtatious conversations with many people, adapting the messages without personally managing every exchange.

That makes it easier to create a false sense of connection and persuade people to click links, pay for content, or share personal or intimate information.

The line between a scripted spam account and a responsive conversational partner is getting harder to see. Judge the account by what it wants you to do, not by how convincingly it talks.

---

Something feel off? Check it before you click.  

Malwarebytes Scam Guard helps you analyze suspicious links, texts, and screenshots instantly.  

Available with Malwarebytes Premium Security for all your devices, and in the Malwarebytes app for iOS and Android.  

Try it free → 

Last week on Malwarebytes Labs:

• The hidden work of modernizing Malwarebytes
• X Money rollout linked to password-reset attacks
• Free streaming boxes may be routing criminal traffic through your home
• StreamRat Android malware spreads through Meta and TikTok ads
• Your phone or computer may soon ask how old you are
• Tech support scams look different now. Here’s what to watch for
• Scammers are getting smarter about where they target you
• Two critical Chrome flaws put users at risk on malicious websites
• 153M+ driver’s licenses for sale on new dark web platform
• Your AI chats could be used in court
• Fake GTA 6 leaked copy drains your crypto wallet
• TerminalFix looks like ClickFix, but delivers a very different payload
• Infostealers are hijacking Claude accounts at users’ expense
• McKesson confirms cyber incident after ShinyHunters claims patient-data theft

Stay safe!

---

Something feel off? Check it before you click.  

Malwarebytes Scam Guard helps you analyze suspicious links, texts, and screenshots instantly.  

Available with Malwarebytes Premium Security for all your devices, and in the Malwarebytes app for iOS and Android.  

Try it free → 

Most of the work that keeps a security product trustworthy is invisible. Users see a scan complete, a threat blocked, an update applied overnight. They don’t see the platform underneath. Runtimes, managed libraries, native drivers, and Windows requirements must all stay current and work together across millions of endpoints. Our migration to .NET 10 is one example of how we keep that platform moving and our focus in this article.

It would be easy to call these upgrades maintenance tasks and move on. But that is underselling them. Our code runs continuously, with elevated privileges, next to some of the most sensitive parts of Windows.

Every dependency in our stack, from the runtime and third-party libraries to native drivers and operating-system requirements, affects the environment in which our software runs. When one changes, everything that relies on it may have to change too.

The problem: platforms fall behind by standing still 

In endpoint security, the ground never stops moving. Windows evolves. Threats evolve. Hardware evolves, from ARM64 laptops to machines with far more memory and faster storage than the ones our code was first written for.  

A platform that stands still does not stay the same. It falls behind. Every skipped release of a dependency or runtime widens the gap between the ecosystem we built on and the one that is stable today.  

A modern runtime and .Net 10 in particular can give us better security, faster code paths, a smaller memory footprint, and richer diagnostics. It also gives our engineers language and tooling improvements that help them work more efficiently. 

The stakes are also particularly high for security software: 

• A web app can be rolled back with a deployment. Software already installed on a customer’s endpoint cannot. 
• Our code runs with high privileges, alongside kernel drivers and anti-tamper protections.   
• A runtime regression does not affect one server. It has the potential to affect millions of machines.  

So we treat a runtime upgrade with the same rigor as a security feature. 

The challenge: everything moves together 

Malwarebytes for Windows is not a single program. It is a coordinated system: a user-facing interface, several long-running Windows services, an installer, a self-update pipeline, a plugin surface, and third-party managed dependencies.  

These sit above native drivers and our detection engine. The .NET 10 migration covered the managed parts of Malwarebytes while leaving this native core untouched. But the different layers still have to work together. 

The migration had to satisfy several requirements at once:  

• Security-sensitive code had to behave identically before and after the change.  
• Native drivers and anti-tamper layers had to continue working correctly with the  managed code. 
• The installer and update pipeline had to deploy the new runtime files and clean up old ones.  
• Plugins and third-party dependencies had to remain compatible. 
• Existing Malwarebytes installations had to continue working.  
• Our automated validation had to be extensive enough to trust the result without inspecting every path by hand. 

The boundary between managed and native code deserved particular attention. Managed code in our services talks to native components through interfaces such as P/Invoke and COM. A runtime change can subtly affect how these different parts of Malwarebytes communicate and work together. 

Those differences may never show up in a demo. They might only surface on one machine in 10,000. Finding them before customers do is the important part. 

Not every update looks the same 

There are three main reasons we update a dependency: We choose to, the platform underneath forces us to, or a vulnerability makes us. Each comes with a different timeline.  

Elective modernization. We may choose to move to a new runtime, a new major version of a library, or a new platform capability to take advantage of new features, fixes, or security improvements.  

Baseline shifts. As platform requirements evolve, some older compatibility constraints can hold back modernization. For example, moving to .NET 10 allowed us to update the application baseline and adopt a newer, supported runtime. As part of the same change, Windows 7 support was deprecated. 

Forced patches. Sometimes a vulnerability is disclosed in a library we ship or a system we depend on. The change is no longer optional and the timeline is not ours. What we can control is our readiness: the testing, release process, and staged rollout that allow us to respond quickly without introducing new problems. 

Different reasons and different timelines, but each requires the same careful approach. 

The .NET migration: Why we did it 

Moving to .NET 10 gives Malwarebytes a more secure, supported, and capable foundation for Windows, with several compounding benefits: 

Security. A modern runtime benefits from Microsoft’s ongoing security work, including safer defaults, stronger cryptography, and mitigations for memory and interoperability bugs. Staying on a runtime that Microsoft actively supports means we can continue to apply fixes when vulnerabilities are discovered.  

A supported foundation. It may not be a glamorous reason, but .NET 10 keeps us on a supported, actively developed platform that is better aligned with newer versions of Windows. It makes it easier to adopt future fixes, features, and improvements as routine work instead of one-off projects. 

Diagnostics and observability. Modern .NET has stronger built-in tracing, metrics, and crash diagnostics. In a security product, understanding how code behaves in the field matters. Better diagnostics help us identify and resolve reliability issues.  

Performance and memory efficiency. Recent .NET releases have improved the just-in-time compiler, garbage collector, and core libraries. Our processes run all day in the background, so how efficiently they run and manage memory matters. These capabilities give us more opportunities to improve efficiency, although the impact will vary between components. 

The .NET migration: How we did it  

The guiding principle is simple: Never advance faster than the evidence allows.  

We started by isolating the work on a dedicated branch. We retargeted the platform and refreshed every managed dependency so the new runtime and the codebase agreed on exactly which components should ship.  

That surfaced some of the less obvious consequences of the upgrade early on: libraries that had been renamed or folded into the runtime, files that were no longer needed, and new ones that had to ship in their place. 

Deployment is easy to overlook and expensive to get wrong. A runtime migration is not only about the code that runs. It is also about what lands on the customer’s disk.  

Our installer and update service had to learn the new runtime’s file layout. They had to remove dependencies the runtime now provides, stop shipping renamed files, and deliver replacements cleanly during both fresh installs and in-place updates.  

Getting deployment right is the difference between an upgrade users never notice and one that creates a support problem. 

Once the build was working correctly, the focus moved to validation. 

The migration involved: 

• Extensive automated testing across services, install, and update paths. 
• Compatibility validation against real-world configurations and previous installations. 
• Performance benchmarking to catch regressions in startup, memory, and scan behavior. 
• Canary and staged deployment, starting with small populations and expanding only when results showed it was safe to do so. 
• Continuous monitoring and automated regression detection in the field. 
• Cross-functional work across platform, QA, installation and update, and release engineering. 

“We never advanced faster than the evidence allowed. Every gate had to turn green on its own.” 

The .NET migration: The tradeoffs  

None of this was free, and the choices involved deliberate tradeoffs.  

The tradeoff on the branch was drift. Isolating the migration protected the main codebase, but the longer that branch existed, the more it could diverge from active development. We managed that risk through frequent integration rather than leaving one large, risky merge until the end.  

The tradeoff on rollout was speed. Staged deployment also meant customers received the update later than they would with a big-bang release. We accepted that tradeoff. Evidence from smaller populations gives us an opportunity to catch problems before an update reaches a much larger number of endpoints. 

The tradeoff on AOT was flexibility. Ahead-of-time compilation can improve startup performance but can also constrain dynamic behavior. We applied it selectively, component by component, rather than everywhere by default. 

What the .NET 10 migration means for customers

The best outcome of infrastructure work is that customers benefit from it without having to think about it.  

On .NET 10, those benefits for Malwarebytes customers include:   

• Improved reliability on a fully supported, actively maintained runtime.  
• A more secure foundation that benefits from the platform’s continuing security improvements.  
• Access to new .NET features and fixes. 
• Better support for newer versions of Windows. 
• Newer .NET tools can help us develop and deliver new protection faster. 

Lessons worth keeping  

Every one of these updates—the runtime, the baselines, the security patches—leaves the team holding a few convictions more firmly.  

Platform upgrades are strategic investments in everything built on top of them. Modernization also works best when it is continuous: the longer a platform falls behind, the more difficult the eventual upgrade can become. 

Automation is particularly important for changes of this breadth. So are the small, unglamorous decisions made years earlier, such as maintaining clean boundaries between components and having a build process that knows precisely what it ships. 

Those foundations are what make larger changes possible.  

“Technical debt compounds like financial debt. The cheapest upgrade is the one you never postponed.”  

What comes next  

No upgrade is a finish line. Each one is a step in a longer pattern: modernize continuously, in deliberate steps, so the platform never falls behind. Baselines will shift again. Vulnerabilities will land without warning. Each will meet the same discipline: the same tests, the same staged rollout, and the same evidence before we move forward. 

That discipline is what a product trusted to run every day, on every machine, without a second thought is actually made of.  

Malwarebytes for Windows on .NET 10 shipped in version 5.6.0. It is the latest step in a long-standing commitment to invest in the platform beneath the product so the protection on top of it can keep getting better. 

---

---

X says attackers may be targeting accounts because its X Money payments service is now more widely available.

The company is investigating a wave of unsolicited password-reset emails sent to users. While their arrival alongside the wider X Money rollout has fueled account-takeover concerns, X says it has found no evidence of a breach or successful account takeovers so far.

X users began reporting unexpected password-reset emails and codes on September 1. In a public post, X product engineer Mridul Singhai said:

“Attackers appear to believe that, now that @XMoney is widely available, they can gain unauthorized access to accounts.”

Singhai said X was actively investigating, apologized for the repeated emails, and added that the company had found “no evidence of any breaches.”

X Money gives eligible US users access to financial services within X, including interest-bearing accounts, a Visa debit card, and peer-to-peer payments. Cross River Bank provides the banking infrastructure behind the service.

This could make some X accounts more attractive targets, particularly accounts with payment access, high follower counts, business use, or valuable social-engineering potential.

The activity itself appears consistent with attackers submitting password-reset requests in bulk against X accounts. Requesting a password reset is not the same as resetting a password, however, and neither automatically means that an account has been taken over. X’s recovery process requires access to the email address or phone number associated with the account before someone can complete the reset.

There is no evidence that anyone has accessed X Money accounts or funds. Nor has X confirmed that X Money caused the password-reset activity. The timing is notable, but it does not prove a technical connection between the two.

Earlier this year, we saw a flood of Instagram password-reset emails, showing that similar activity can happen on platforms without payment services.

It could be a cover for something more serious. Even if an attacker cannot complete a reset, large volumes of legitimate-looking reset messages can provide useful cover for scams.

Reset flooding can also be a nuisance tactic. Repeated messages may pressure someone into changing their password unnecessarily, obscure more important security notifications, or encourage them to disable security controls in an attempt to stop the alerts.

How to stay safe

If you receive an X password-reset email that you did not request:

• Do not click links or enter codes from unexpected messages. Open the X app or type x.com into your browser yourself if you want to inspect or change account settings.
• Do not share reset codes or two-factor authentication codes. Support staff, advertisers, and “security teams” will not contact you unexpectedly to ask for them.
• Turn on password-reset protection. X says this setting requires additional account information, such as an email address or phone number, before it will send a reset link or code. It is available under Settings and privacy > Account > Security > Password reset protection.
• Use two-factor authentication, preferably an authenticator app or security key where available. This adds another verification step if someone obtains or guesses your password.
• Use a unique, strong password. If you use your X password anywhere else, change it through X’s settings, not through a link in an email.
• Watch for signs of an actual account takeover. These include unfamiliar posts, direct messages, profile changes, login alerts, or unknown apps connected to your account.
• Stay alert for phishing. The strongest immediate consumer risk may not be a flaw in X itself, but phishing that imitates the reset process. A fake message can look especially convincing when genuine reset emails are arriving around the same time.
• Use protection. An up-to-date, real-time anti-malware solution with web protection can warn you about malicious and fraudulent sites.

If you’re unsure whether a message is real, use Malwarebytes Scam Guard to check it and get advice about what to do next.

---

Scammers don’t need to hack you. They just need you to click once. 

Malwarebytes Identity Theft Protection catches suspicious activity before it becomes a problem.

“Free” movies and TV could cost you your privacy, bandwidth, and control of your home network.

We’ve warned about illegal streaming and modded Amazon Fire TV Sticks in the past. Now, researchers have found that certain SuperBox devices and apps could quietly enroll a household connection into a proxy network, allowing third parties to route traffic through it.

An earlier report identified CyberFlix TV, available through SuperBox’s custom app store, as containing Popanet proxy functionality that registers the device with a server controlled by the proxy operator.

Law enforcement agencies have warned that “foreign entities” are using residential proxies to conceal their identities and make their activity appear to come from someone else’s home network.

The FBI defines a residential proxy as follows:

“A residential proxy is an intermediary server between individuals and websites they visit to make their connections appear to originate elsewhere. Legitimate IP addresses assigned by an Internet Service Provider (ISP) to consumers’ Internet of Things (IoT) devices, such as TV streaming devices, digital picture frames, smartphones, tablets, and routers are used to route traffic. Once an internet-connected device is compromised, the device’s IP address can be used by threat actors to mask their online illegal activity, making the consumer appear responsible.”

Residential proxy networks rent out ordinary home IP addresses to customers. That makes their traffic appear as if it originates from a legitimate consumer connection rather than a data center, helping cybercriminals evade IP-based fraud controls and reputation systems.

Besides affecting connectivity, this can mean that a household’s public IP address becomes associated with activity it did not initiate, ranging from credential stuffing and account abuse to attempts to bypass enterprise security controls.

Plume’s more recent research warns that these proxy networks can also function as malware-delivery platforms. In other words, attackers may not only use a compromised streaming box as an exit node. They may use the proxy connection to reach the box itself and install additional malicious software.

The reported SuperBox configuration is especially troubling because it disables or weakens multiple Android safeguards. Researchers found exposed Android Debug Bridge (ADB) access, root-level privileges without authentication, and the removal of protections that would normally restrict untrusted app installation or prompt users to approve risky actions.

Many people assume that placing a streaming device behind a home router prevents outside access. Normally, network address translation and a firewall do make unsolicited inbound connections more difficult. But proxy-enabled devices can maintain an encrypted outbound connection to a remote server, creating a channel that the home router treats as legitimate traffic initiated from inside the network.

How to stay safe

The safest option is not to connect devices or install apps that promise unauthorized access to free movies and TV. You could be bringing a proverbial Trojan horse into your home.

If you own a SuperBox device or have installed CyberFlix TV, disconnect the device from your network. A factory reset may not be enough to make it safe to use again, so you should replace it.

When a device’s business model depends on monetizing your connection, its security choices can put your IP address, bandwidth, privacy, and local network at risk.

Network segmentation can reduce exposure for ordinary Internet of Things (IoT) devices, but it is not a complete answer here. A product that intentionally establishes a persistent proxy channel and offers weak device-level protection should not be trusted on a household network, even on a separate guest network.

---

From reporting threats to removing them.

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

A malicious advertising campaign promoting a fake free TV-streaming service reached roughly 570,000 Meta users.

The researchers who discovered the campaign found that its streaming-themed ads were aimed at Spanish-speaking users, with most observed victims located in Spain. One Meta campaign ran from June 11 through July 3, 2026, and the same banners were also used to distribute the malware through TikTok.

The available data shows the ads’ reach, not the number of downloads or infections, but it demonstrates how quickly paid advertising can put a scam in front of a very large audience.

The ads promoted an Android banking Trojan and infostealer called StreamRat. It can monitor what’s on screen, capture information typed into apps, show convincing fake screens to steal usernames and passwords, and allow attackers to control the device remotely.

We often warn people not to click suspicious links in unexpected texts or emails. But malicious advertising is harder to recognize because it appears in the same feeds where people expect to find promotions, videos, and recommendations.

This campaign is a perfect demonstration of why “after-the-fact” ad checks are inadequate when it comes to protecting social media users. Attackers used familiar social media advertising and carefully tailored instructions to turn casual interest in free entertainment into a risky app installation.

How the attack worked

The ad led victims to a website posing as a streaming platform. The site checked whether a visitor was using Android. Non-Android visitors were simply prevented from downloading anything, while Android users were shown an app download option. This is a common way for scammers to concentrate their efforts on devices their malware can infect.

The site also identified whether someone had arrived through Instagram, TikTok, Facebook, or a regular browser. It then displayed instructions suited to that situation, including steps to allow the browser to install apps from “unknown sources.” In other words, this was not a generic malicious download page: It was designed to coach people through the security warnings that would normally make them stop and think.

StreamRat is an Android banking Trojan and infostealer. It can monitor what’s on screen, capture information typed into apps, show convincing fake screens to collect usernames and passwords, and enable attackers to operate the device remotely. The researchers also found options to cover the screen with a black page or fake Android update screen. These can distract victims while criminals interact with the phone behind the scenes.

How to stay safe

While this campaign targeted Spanish-speaking people, primarily in Spain, the following guidelines can help anyone avoid similar attacks.

• Avoid installing Android apps from ads, direct-download websites, social media messages, sponsored search results, or links sent by strangers.
• Download apps through Google Play whenever possible, and check the developer’s name, reviews, and app history rather than relying on an ad.
• Before enabling installation from “unknown sources,” read our guide, Sideloading on Android: What it is, why it’s risky, and how to do it more safely.
• Be very cautious when an app asks for Accessibility access, screen-sharing permission, Device Admin privileges, or permission to become the default launcher. Permissions that don’t line up with the intended use of the app are very suspicious.
• Use an up-to-date, real-time anti-malware solution on all your devices.

What to do if you installed a suspicious app

If you installed a suspicious APK and granted it Accessibility access, disconnect the phone from Wi-Fi and mobile data. If possible, revoke the app’s Accessibility access and remove it. Use another device to change relevant passwords and contact your bank if you used banking apps on the infected phone. A factory reset may be necessary if you cannot confidently remove the infection.

Malwarebytes for Android detects the components of StreamRat as Android/Trojan.Agent.ACRAEEF8A36H36, Android/Trojan.Agent.ACR02DB0614H7, and Android/Trojan.Dropper.ACR9B7ECE83D1.

---

Scammers know more about you than you think. 

Malwarebytes Mobile Security protects you from phishing, scam texts, malicious sites, and more. With real-time AI-powered Scam Guard built right in. 

Download for iOS → Download for Android → 

First, the good news: If you use a Linux-based operating system, you may not be asked your age in a few months. The bad news is that Windows, macOS, iOS, and Android users in California will be.

California has passed a law that requires a range of operating systems to start collecting your age when you first set them up. Under the state’s Digital Age Assurance Act (DAAA), signed into law in October 2025, Windows, macOS, iOS, and Android will all have to do this from January 1, 2027.  Operating systems set up before that date in California will need to do the same by July 1, 2027.

Operating systems will categorize people into four age brackets: under 13, 13–15, 16–17, and 18+. They will then be able to send a non-identifying age signal to app developers. Developers must request that signal from the operating system provider or app store when someone downloads and launches an app. This makes them legally aware of the person’s age bracket.

California wants to stop children from doing things that could hurt them. Kids shouldn’t be able to download apps containing mature content meant only for adults, for example. Age assurance also goes hand in hand with social media restrictions, and Meta recently agreed to put time limits on kids’ social network use as part of a massive court settlement. Another California bill, AB1709, would restrict addictive social media features for children under 16. Measures like these need some form of age assurance to function.

This makes digital rights activists unhappy. The Electronic Frontier Foundation (EFF) isn’t a fan of age verification. It accused California of “outsourcing censorship to developers” through the DAAA rather than focusing on privacy.

The EFF was also uncomfortable with the effect of all this on open-source systems. Age verification requires time and effort from operating system developers. That’s fine if you’re Microsoft, Apple, or Google with a massive development budget. But it’s more problematic for operating systems developed by volunteers, such as Linux distributions. Those that don’t have the resources to comply, or don’t like the privacy implications, might prefer to avoid the Golden State altogether.

GrapheneOS, a privacy-focused mobile operating system that strips Android of its surveillance functions, took that option. In March, it said that it wouldn’t implement age verification, and would happily forego sales of devices running its software in certain regions, if necessary.

Assembly member Buffy Wicks, who introduced the original DAAA, has been listening. She tweaked the legislation with Bill AB1856, which would amend the law to exempt certain open-source operating system providers. California lawmakers passed the bill in late August, and it is now awaiting the governor’s decision.

AB1856 would exempt software that follows open-source rules, allowing it to be reused and built upon by others. This includes software distributed under common licenses such as GPL, MIT, BSD, and Apache. Not one single lawmaker voted against it.

California isn’t alone in mandating the collection of age brackets. Colorado’s SB26-051, now law, does something similar. Legislators there also added parallel open-source exemptions after lobbying by Linux hardware maker System76.  Illinois has also passed age assurance legislation, and New York has a bill in the works.

Exempting open-source operating systems from California and Colorado will please privacy-conscious users, but it’s worth noting that some Linux distributions are going ahead with age assurance anyway. Many have drawn a line in the sand, others, like Fedora, are reportedly planning to do it anyway.

In any case, those using more mainstream operating systems can expect a “How old are you?” or “What’s your birthdate?” question sometime soon. If you’d rather avoid that, consider an open-source operating system instead. Just check with your distribution’s maintainers to see what their plans are.

---

From reporting threats to removing them.

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

In a tech support scam, criminals pretend to work for a trusted technology or security company. They claim there is a problem with your device, software, subscription, or account, then try to persuade you to pay them, share personal information, or give them remote access to your computer.

These scams used to rely mainly on browser locks and fake virus warnings. Now, scammers use many more ways to reach people, including websites and platforms they trust.

How tech support scams reach you

As well as copying the websites of reputable brands, tech support scammers abuse sponsored search results, hijack on-site searches, create fake listings on trusted platforms, and use renewal scams, fake calendar invites, Apple Pay notifications, and many other methods to persuade people to call them.

Once someone makes contact, the scammers may demand payment, ask for personal information, or try to persuade them to install remote access software.

Beware of someone wanting to connect to your computer remotely. One of a tech support scammer’s most powerful weapons is the ability to connect remotely to a victim’s computer. If you allow this, the scammer may gain access to all of your files, folders, and the information they contain. 

Tech support scams impersonating Malwarebytes

Tech support scams affect Malwarebytes directly because scammers often impersonate trusted security companies, as in the example below.

You can tell it’s not the real Malwarebytes when:

• They use a name other than Malwarebytes. Malwarebytes does not outsource its support. We have our own Support team and do not authorize third parties to provide support using our name, logo, or any other intellectual property. 
• They can’t or won’t accept payment by credit card. Malwarebytes uses a credit card processor for all transactions. Credit card processors screen the companies they work with for risks such as fraud and abuse. Credit cards also offer consumer fraud protections, so it is a red flag if a company tries to steer you toward another payment method.
• They make unsolicited support calls. Malwarebytes does not do this. Tech support scammers may buy personal information from data brokers that have identified people as potentially vulnerable targets. But how would a legitimate company know that you have a problem with your computer—or even that you own one? If someone calls out of the blue claiming that your computer has a problem, hang up.

What to do if you’ve been scammed

If you’ve fallen victim to a tech support scam, here are a few steps you can take:

• Have you already paid? Contact your credit card company or bank and let them know what’s happened. You may also need to file a complaint with the FTC or contact your local law enforcement agency, depending on your region.
• Did you share your password with the scammer? Change it on every account that uses the same password. Consider using a password manager and enabling two-factor authentication (2FA) on important accounts.
• Scan your system. If scammers have accessed your computer, they may have installed a backdoor that allows them to return later. Malwarebytes can remove backdoors and other software left behind by scammers.
• Keep an eye out for unexpected payments. Look for suspicious charges or payments on your credit cards and bank accounts so you can dispute them quickly and prevent further losses.
• Be wary of suspicious emails and text messages. Scammers may now see you as a potential target and try other methods to defraud you.

How Malwarebytes is fighting tech support scams

Malwarebytes researchers actively fight tech support scams in the US and overseas. They work closely with the Federal Trade Commission (FTC), providing technical evidence to help shut down tech support scammers and educating internet users about the latest tactics and how to protect themselves.

Malwarebytes is also a supporting member of the Global Anti-Scam Alliance (GASA), working with other organizations committed to reducing scams and keeping people safer online.

---

Something feel off? Check it before you click.  

Malwarebytes Scam Guard helps you analyze suspicious links, texts, and screenshots instantly.  

Available with Malwarebytes Premium Security for all your devices, and in the Malwarebytes app for iOS and Android.  

Try it free → 

Scammers are becoming more strategic about where they target people. 

Nine in ten toll scams—the fake unpaid-toll messages that threaten fines or license suspension—arrive by email or text, while roughly six in ten romance scams show up first on social media. That’s no coincidence. Rather than blasting the same message everywhere, criminals are tailoring different scams to the platforms where they’re most likely to succeed. 

This finding comes from Malwarebytes’ own threat research systems and draws on global data between April 15 and July 14, 2026. The research reveals the various ways scammers are adapting their tactics and provides new insights about where they show up, when they strike, and which brands they impersonate.  

Here’s a look at the key takeaways. 

Every scam has a preferred platform 

You’re far more likely to receive a fake giveaway scam via a social media feed than you are by email or text. On the other hand, half of all IRS scams will come via a phone call. Malwarebytes measured more than 20 different types of scams, ranging from tech support and refund scams to sextortion and scareware, and found that each one favors a specific platform.

Intuitively, the platforms favored often match the content of a scam—job scams mostly arrive through typical work channels like email, romance scams mostly arrive through social media where meeting strangers is least questioned, and tech support scams mostly arrive through the phone. The channel can also shape how the scam feels: A DM can feel personal, while a phone call creates pressure to respond in the moment. And prior research shows that scammers often repeat what works, which might explain why they keep doing what they’ve been doing so far.  

The web still wins 

Despite the rise of social media and messaging apps, scams are reaching us through the web more than any other platform, followed by email and SMS.  

No surprise that the web is the most popular doorway. Malwarebytes blocks around 500,000 phishing websites a day.  

MrBeast beats Trump 

He’s already the most popular YouTuber in the world, famous for his online antics and extreme stunts, but Malwarebytes data shows that “MrBeast” can now add “most impersonated person” to his resume, handily beating Elon Musk and Donald Trump (numbers two and three, respectively).

MrBeast, whose real name is Jimmy Donaldson, is the go-to favorite for scammers looking to piggyback on his fame by using his likeness in some 30% of impersonation scams, ranging from crypto giveaways to transfer fee swindles. Familiar faces lower the public’s guard and make scam messages feel more credible, which is why they work so well. So, the next time MrBeast shows up in your feed asking you to send cash as part of a verification process, watch out.  

12:00 pm ET is the “golden hour” for scammers targeting Americans 

If you’re on the East Coast in America, lunchtime is also scamming time. Malwarebytes data shows that high noon is the golden hour for scam texts, and it’s roughly 874% busier than the quietest time, which is 1:00 am ET.    

Scam texts peak on Fridays 

The rate of scam texts hitting your phone builds throughout the week. From a low on Sunday, they increase steadily in frequency and hit their peak on Fridays. So by the time you’re leaving work and preparing for the weekend, you’re also getting hit with roughly 50% more fraudulent text messages than you were when the week began. The data doesn’t tell us why, but it does suggest scammers are deliberately timing their campaigns rather than sending messages at random. 

Big brands are big business for scammers  

The world’s biggest brands are also some of the most useful to scammers. They’re instantly recognizable, used by hundreds of millions of people, and already part of our everyday lives, making them a natural fit for everything from fake offers to bogus account alerts. Based on reports from Malwarebytes users, the five most impersonated brands are: 

• Google
• Microsoft 
• Apple 
• Roblox 
• Amazon 

According to Malwarebytes users, Google’s brand name was abused at least twice as often as Amazon.  

Gaming is becoming a bigger target 

Scammers are increasingly targeting gaming communities. According to data collected by Malwarebytes Scam Guard, about half of all gaming scams can lead to a financial hit of $1,000 or more, what we term a “high-severity risk.” The most impersonated gaming sites were Roblox, Steam, Discord, and Minecraft. Roblox saw a 15% spike in scam activity from mid-June to mid-July, while Steam saw a 19% spike over the same period. 

How to spot and stop scams 

The data points to a scam economy that’s becoming more specialized. Rather than relying on one-size-fits-all campaigns, criminals are tailoring scams to the platforms we use every day, from email and text messages to gaming communities and social media. The tactics may change, but the goal stays the same: to earn your trust long enough to steal your money or your information. Knowing how those tactics vary from platform to platform makes them easier to recognize, and easier to avoid. 

In general:  

• Do not click links or call phone numbers in unsolicited emails, text messages, or social media DMs. 
• When in doubt, check the legitimacy of the message by going directly to the company’s official website and asking about it through official channels. Don’t follow sponsored search results to get there—these can be scams. 
• Do not give out personal details, PINs, passwords, payment information, or verification codes during an unsolicited call. Legitimate companies will not ask for passwords or verification codes over the phone. Hang up and call back through the organization’s official phone number. 
• Use a browser extension that blocks scam and phishing sites, such as Malwarebytes Browser Guard. It can flag a fake storefront before you land on it, including ones it hasn’t seen before. 
• Install a mobile security product like Malwarebytes Mobile Security that filters out scam and spam text messages. 
• Check if a message is legitimate. Malwarebytes Scam Guard can check a message, phone number, or link against its expansive threat intelligence database to determine if it is malicious or safe. It then provides information on red flags and any next steps you should take. Scam Guard flags approximately one in five analyzed sessions as high-risk—situations that could result in significant financial losses ($1,000 or more) or personal harm. 

Methodology   

The data in this report comes from Malwarebytes proprietary threat research systems, collected between April 15 and July 14, 2026. All data is anonymized and reflects what Malwarebytes is able to observe through its own infrastructure. 

This report is also available to read in PDF format.

Chrome is rolling out an update for its desktop browser. The update includes 26 security fixes, two of which Google rates as critical use-after-free vulnerabilities.

How to update Chrome

If you don’t want to wait for the rollout to reach you, manually updating is easy.

The easiest option is to allow Chrome to update automatically. But you can end up lagging behind if you never close your browser or if something goes wrong with the update.

To update manually, click the More menu (three dots), then go to Settings > About Chrome. If an update is available, Chrome will start downloading it automatically. Restart Chrome to complete the update, and you’ll be protected against these vulnerabilities.

You can find an explanation of the version numbering system and step-by-step instructions in our guide: How to update Chrome on every operating system.

Technical details

Let’s look at the two critical vulnerabilities. Both are use-after-free (UAF) vulnerabilities. A use-after-free vulnerability occurs when a program attempts to access a memory location after it has been freed. That can cause crashes or, in some cases, allow an attacker to run their own code.

The first, tracked as CVE-2026-84353, was found in Shared Tab Groups and could allow a remote attacker using social engineering to execute arbitrary code outside the browser sandbox via a crafted HTML page. Here, social engineering likely means an attacker would have to lure you to a malicious website or open an email in HTML format.

The other critical vulnerability, tracked as CVE-2026-84352, was found in WebGL. WebGL, short for Web Graphics Library, is a browser technology that lets websites display interactive 2D and 3D graphics. The vulnerability could allow a remote attacker to execute arbitrary code outside the sandbox via a crafted HTML page.

Chrome vulnerabilities that enable remote code execution outside the browser sandbox are particularly valuable to attackers because they can turn a visit to a malicious or compromised website into direct code running on the underlying operating system, often without requiring additional exploitation steps.

---

Stop threats before they can do any harm.

Malwarebytes Browser Guard blocks phishing pages and malicious sites automatically. Free, one click to install. Add it to your browser →

A new dark web platform called Nexus claimed to be selling 153 million driver’s license scans and millions of other identity and medical cards.

The collection included more than 153 million driver’s licenses, 10 million ID cards, 3 million travel documents, and 579,000 medical cards, including marijuana dispensary cards, according to reports.

The trove of driver’s license scans reported by KrebsOnSecurity is a sharp reminder that identity verification is not a harmless box-ticking exercise.

The FBI’s New Orleans field office has opened an investigation into an apparent breach involving identity verification provider IDScan.net. The company said it was investigating.

IDScan.net advertises as follows:

“We provide simple, secure solutions to help dispensaries reduce liability and protect their licenses by validating IDs, including a customer’s age, in a matter of seconds.”

The allegedly exposed records were especially concerning because some included more than a basic photo of an ID. KrebsOnSecurity found records containing front-and-back images, as well as infrared and ultraviolet scans, with timestamps that appeared to align with the holders’ travel or car-rental activity.

That matters because a driver’s license is far more useful to an identity thief than a password. You can reset a password. You cannot easily replace your face, date of birth, address, or license number, particularly when they’re accompanied by high-resolution images of your government-issued ID.

The age-verification problem

Age verification has become a common justification for asking people to upload an ID, take a selfie, or submit both to a third-party identity verification provider.

We have previously warned about the privacy and security trade-offs in age-verification systems, particularly those that require people to submit copies of government-issued ID. Such systems can turn a request to access a website into a decision to share an enduring identity document with a company the user may never have heard of.

In our opinion, that is a disproportionate risk. Once someone uploads an ID, the service or its vendor can potentially link the visit to their identity. If the provider is breached, the consequences can extend well beyond unwanted marketing or an exposed email address.

The reported Nexus dataset illustrates a broader concern: Identity documents are collected in many places that people may not connect with one another. Each individual collection may be presented as routine, but together they create an ever-expanding ecosystem of organizations, contractors, software platforms, cloud services, and privacy policies.

Facial images and ID copies can be reused. Criminals may use them to make scams more convincing, pass weak identity checks, or assemble detailed victim profiles from records obtained from separate breaches. An attacker who knows your name, address, date of birth, email address, and license details has a useful foundation for fraud.

This is why “we only need to verify your age” should not automatically mean “please upload your driver’s license” or another form of ID.

How to stay safe

When an ID check is required to use an online service, ask a basic question: Why does this company need a copy of my identity document, and what happens to it afterward? The scale of the data reportedly offered through Nexus shows why the answer matters.

Consumers cannot always refuse an ID check, particularly where it is legally required or necessary for a regulated service. But you can reduce unnecessary exposure:

• Ask whether an ID image is stored and, if so, for how long.
• Check whether the company uses a third-party identity verification provider.
• Prefer services that offer a privacy-preserving age check rather than requiring a full ID upload.
• Avoid submitting identity documents to sites you do not trust or did not intend to use.
• Do not email copies of IDs unless there is no safer alternative and you have independently verified the recipient.
• Be alert for phishing, account-recovery scams, and fraudulent credit applications if you believe your ID may have been exposed.
• Consider a credit freeze where available.

---

Let’s face it, an incognito window can only do so much. 
 
Breaches, dark web trading, credit fraud. Malwarebytes Identity Theft Protection monitors for all of it, alerts you fast, and comes with identity theft insurance. 

You might tell an AI chatbot secrets that you wouldn’t divulge to your closest friends. If you do, though, beware: They could end up as evidence in court.

An article in the Washington Post this week highlighted several cases in which people had discussed sensitive information with AI systems like Claude and ChatGPT, only to have their conversations obtained by prosecutors or opposing lawyers.

Lawyers can get access to your chatbot conversations from AI services like ChatGPT because they aren’t privileged in the same way that, say, a conversation with your lawyer or a doctor would be.

Reporters at the paper found chatbot logs cited in 12 court cases in the past two years. They also found statistics from OpenAI that supported a rising trend in data disclosures. The company, which operates ChatGPT, disclosed the content of more than 80 user accounts in the last six months of 2025. That was more than four times as many as in the second half of 2024.

Cases are piling up

With people asking AI for all kinds of advice, it’s little wonder that lawyers are coming after that data too. Sometimes, it emerges because users consent to a search. The Washington Post mentions one university student who asked ChatGPT in a panic whether people might work out that he had damaged 17 cars in a campus parking lot. He then handed his phone over to police for a search. A teen suing big tech companies over social media addiction saw his own ChatGPT history drawn into discovery.

Deleting your chats isn’t watertight protection either. In The New York Times’ copyright lawsuit against OpenAI over collecting its content for training data, a judge ordered the AI company to preserve chat logs, including ones that users had asked it to erase. OpenAI complained that users were being “forced to forgo the privacy protections OpenAI has painstakingly put in place.” The company had to keep that data even though it had agreed to delete it under the EU’s General Data Protection Regulation (GDPR) and California’s privacy laws.

Incidents like these involve responses to legal requests, but AI companies don’t always wait for a subpoena. OpenAI’s policy allows its reviewers to refer conversations to law enforcement whenever they identify “an imminent and credible risk of harm to others.” The Washington Post reported an incident in which OpenAI contacted police after a ChatGPT user in Palm Beach County, Florida, repeatedly described plans to harm an ex-girlfriend.

Technology companies have been handing over all kinds of data beyond AI chats to law enforcement and litigants for years. Google, Meta, and Apple shared details of 3.16 million US user accounts between 2014 and 2024, with substantial increases in the number of records shared annually during that period.

Every time a new technology emerges, litigants will go after it for data. In 2019, police issued a subpoena for audio recordings from an Amazon Echo owned by a Florida man charged with murdering his girlfriend.

What to do

We’d all like to think that true friends will carry our secrets to the grave. But AI is not your friend. Or your doctor, or your lawyer. Treat all chats as records that could potentially be disclosed in a legal case. They might feel like informal conversations, but you should assume that each one creates a written record, even if there’s a delete button.

Be careful about what you share. If the topic is one you’d normally raise only with a doctor or a lawyer, then raise it with a doctor or a lawyer, not AI. Communications with lawyers may be protected by attorney-client privilege, while medical information is subject to confidentiality and privacy protections. Chatbot conversations aren’t.

Finally, be cautious beyond AI. Everything from ill-advised social media posts to private messages might also find its way into police hands. In 2022, for example, Facebook handed over private messages between a mother and daughter to police investigating an illegal abortion case.

So think twice before posting anything sensitive.

---

From reporting threats to removing them.

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

We’ve seen scam sites built around Grand Theft Auto VI (GTA 6) targeting visitors in three different ways this year.

In June, we looked at sites selling GTA 6 “early access” for hundreds of dollars in cryptocurrency. You paid, got nothing, and could not reverse the payment. In August, we found fake Extended Look and demo sites delivering an infostealer instead of a game.

The site we examined this week looks like a GTA 6 fan countdown site but offers to sell a leaked copy of the game. It loads a wallet drainer the moment you arrive: code designed to steal cryptocurrency and other assets from connected wallets. Choose to pay with cryptocurrency, and the drainer asks you to connect a wallet. It can target assets across several blockchain networks.

What the page looks like

The disguise works because much of the page uses accurate-looking information. There is a live countdown to November 19, a map of Leonida, and a grid of release facts and gameplay tiles. The release details match Rockstar’s own announcements: GTA 6 is scheduled for November 19 on PlayStation 5 and Xbox Series X|S, and Rockstar has not announced a PC version.

Two offers sit among that accurate material. One sells a leaked copy for $50. The other offers the same thing for cryptocurrency: 1 SOL (worth about $102 at the time of writing).

The page then tells visitors that every other site offering leaked material is a scam and this one is the only safe place to buy. Warning visitors about other scammers is a common technique designed to reassure anyone who is already suspicious.

The site also contradicts itself in ways anyone can check. Its footer states that the page offers no purchase, download, or payment of any kind, directly beneath two payment buttons. The signup box is headed “Get notified. Not scammed.” The facts grid says the game is console-only, while the FAQ promises a PC download after purchase, and claims no price has been confirmed, months after Rockstar opened pre-orders in June.

The writing splits in two as well. The countdown and map sections are clean copy. The sales copy contains multiple errors, including a misspelled “download” and a reference to GTA IV rather than VI. Our reading is that a legitimate-looking fan template was reused and the sales pitch added by someone else who did not proofread it.

What happens when you connect a wallet

There are two pieces of code here, and they are very different.

The first is written into the page and targets a Solana wallet. It does not charge the advertised price. Instead, it checks the wallet’s balance, leaves a small amount to cover the transaction fee, and prepares to transfer everything else to the attacker. The advertised price plays no part in the calculation.

The second is a separate script of around 2.4 MB, and it is far more capable. It includes a legitimate, widely used tool for connecting websites to cryptocurrency wallets, allowing it to work with many wallets rather than just one. Added to that tool is malicious code that inventories the connected wallet, calculates what its assets are worth, reports the details to the attacker, and retrieves transactions for the victim to approve.

The script is configured to target wallets across seven blockchain networks: Ethereum, Polygon, BNB Smart Chain, Avalanche, Arbitrum, Base, and Fantom. It recognizes major stablecoins on those networks and can request several kinds of access. Depending on what the victim approves, it could transfer cryptocurrency immediately or gain permission to move tokens and entire NFT collections later.

A transfer takes assets immediately. An approval can give the attacker access to them later. This script supports both.

It checks where you are first

Before asking the visitor to connect a wallet, the script downloads its settings from the operator’s server. If a particular setting is enabled, it uses the visitor’s IP address to identify their country and checks it against a fixed list: Armenia, Azerbaijan, Belarus, Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan, Turkmenistan, and Uzbekistan.

Visitors from those countries see “This website is unavailable in your region” and are redirected to a blank page. Everyone else continues.

The setting that controls this country-blocking feature is named CIS_Protection in the code.

Excluding this group of countries is a long-standing convention in some criminal tooling, usually interpreted as an attempt to avoid local law enforcement. We would not draw conclusions about who is behind this from a country list alone. What it shows is that whoever built the tool made a deliberate decision about who they were willing to rob and wrote that decision into a settings file.

It works out what your wallet is worth

The script profiles visitors before asking them to approve anything. It checks their holdings across different blockchains, calculates their total value, and sends the details to the operator. These include the wallet’s estimated dollar value, its tokens and NFTs, the visitor’s IP address and country, and how many times the wallet has connected.

The script is also designed to make analysis more difficult. It can detect the automated browsers used by security scanners, suppress messages that would normally appear in the browser’s developer console, interfere with developer tools, and conceal its server addresses inside the code.

Several details suggest that the drainer is rented rather than homemade. The Solana address written into the webpage does not appear in the larger script. Instead, the script downloads an operator ID and settings from a remote server, which also prepares the transactions shown to victims. This resembles a hosted service used by multiple customers, although we cannot identify the product. It also allows the destination of stolen funds to be changed without altering the website.

The fake GTA 6 sites we investigated in June asked victims to send a fixed payment, limiting the immediate loss to that payment. This site can try to take everything in the connected wallet. Depending on what the victim approves, the attacker could either transfer the wallet’s current balance immediately or gain permission to take tokens and NFTs later.

What to look for before you approve anything

Simply connecting your wallet does not allow the site to take anything. The danger comes when you approve the transaction or permission request that follows.

That approval screen is an important last line of defense. The Phantom crypto wallet, for instance, says that it simulates every transaction before you sign and shows a plain-language preview of what will happen, including a warning if something looks suspicious. Other reputable wallets do the same, but they cannot protect you if you approve a request without reading it.

Check the wallet’s approval screen for two warning signs. First, reject any transaction that would transfer all or nearly all of your balance instead of the price you expected to pay. Second, reject any request to approve, allow, or grant access to your tokens or NFTs. That could let the attacker move those assets later, and a shop selling a game has no reason to request such permission.

How to protect yourself

• Nobody is selling a playable copy of GTA 6 yet. Rockstar is selling pre-orders for a November 19 release. Any site offering a leaked, early, or playable copy is not an authorized seller.
• Treat a wallet connection request on a game site as a stop sign. Legitimate GTA 6 purchases are available through Rockstar’s authorized stores and retailers. Rockstar does not ask buyers to connect a cryptocurrency wallet or send cryptocurrency to a wallet address.
• Read the approval screen every time, and reject anything that moves close to your entire balance or asks for ongoing access.
• Don’t let the accurate parts vouch for the rest. A correct release date and real artwork cost an attacker nothing.
• Keep large balances out of the wallet you browse with.
• Block the pages before they load. Malwarebytes Browser Guard is free and blocks scam and malicious sites while you browse.

What to do if you connected a wallet

1. Review and revoke any permissions granted through the site. These permissions may allow the attacker to take assets later, even if nothing has been stolen yet.
2. Disconnect the site from your wallet to end the current connection. This does not cancel any permissions you have already granted.
3. Check the wallet’s full contents, including tokens and NFTs on every chain you use.
4. If funds have been taken or you entered your recovery phrase, move anything of value that remains to a newly created wallet.
5. Report the receiving address to your wallet provider and a public scam-reporting service. This may help providers identify the address and warn other users.

Be wary of anyone offering to recover stolen cryptocurrency for a fee. This is often a second scam aimed at the same victim.

A completed transfer cannot be reversed. The code we analyzed does not request or expose the wallet’s recovery phrase, so connecting to the site alone doesn’t compromise that phrase. If you entered it anywhere during the process, treat that as a separate compromise and move your remaining assets to a newly created wallet. Any permissions you granted remain active until you revoke them.

Remember

The lure has not changed since June. It’s still a promise to play GTA 6 before Rockstar releases it, and that promise is still impossible to keep.

What has changed is what sits behind it. One approach charged a price. Another stole passwords. This one asks for approval to access wallets using a tool built to be rented, reconfigured, and pointed at whatever people are excited about next.

GTA 6 is scheduled to arrive on November 19, 2026, through the same stores gamers already use. No unauthorized playable copy before launch should be treated as legitimate.

Indicators of compromise (IOCs)

Drainer infrastructure

• centrodigestionedellarapina[.]life 
• dasunerforschtelandamendederwelt[.]sbs

Solana address used by the page’s inline transfer

21iWU6FJWJ9FKKz4Jek2CyTh2x1fqs5jawjrNgE3nHjN

Microsoft has published details about a Windows malware campaign it calls TerminalFix.

The social engineering used to infect people is very similar to what we’ve seen in ClickFix campaigns. A website visitor is presented with a fake Cloudflare CAPTCHA which, when clicked, secretly copies a malicious command to their clipboard.

Then they receive instructions on what they need to do to “prove they are human,” when in reality they are being instructed to execute the malicious command.

After gaining a foothold, the malware downloads what appear to be ordinary PNG images from attacker-controlled sites, but the images also contain concealed payload data. A PowerShell script reads the images’ pixel data: The first eight bytes specify the embedded file’s length, while the remaining data is reconstructed into an executable and two DLL fragments, which are joined on the victim’s disk.

The use of steganography is notable, but does not set TerminalFix apart from other ClickFix campaigns that have used the same method.

What does make TerminalFix different is its payload. One way or another, ClickFix victims usually end up with information-stealing malware of some sort.

TerminalFix instead chains together several mature evasion techniques, including DLL sideloading, steganographic delivery, folder hiding, realistic browser User-Agent rotation, and encrypted WebSocket traffic, rather than relying on one obfuscated PowerShell downloader. Its end goal is also different.

The campaign delivers a payload chain that performs domain-aware reconnaissance before installing a custom, multiplexed reverse TCP tunnel.

This hidden connection can give attackers access to the victim’s network. The infected computer first calls out to the attackers using encrypted web-like traffic over port 443, allowing the “reverse” connection to pass more easily through a company firewall.

In this way, the compromised system creates a secret remote-access doorway from inside the network. “Multiplexed” means the attackers can send several separate connections, for example to a file server, database, or another workstation, through that single encrypted tunnel at the same time. This allows them to use the victim’s computer as a stepping stone into the rest of the network.

How to stay safe

First, remember that a real CAPTCHA may ask you to click boxes or select images, but it will not ask you to open Run, Terminal, Command Prompt, or PowerShell and paste a command.

• Slow down. Don’t rush to follow instructions on a webpage or prompt, especially if it asks you to run commands on your device or copy-paste code. Attackers rely on urgency to discourage careful thought, so be cautious of pages urging immediate action. Sophisticated ClickFix pages may add countdowns, user counters, or other pressure tactics to make you act quickly.
• Avoid running commands or scripts from untrusted sources. Never run code or commands copied from websites, emails, or messages unless you trust the source and understand exactly what the command will do.
• Verify instructions independently. If a website tells you to execute a command or perform a technical action, check through official documentation or contact support before proceeding.
• Limit the use of copy and paste for commands. Manually typing commands instead of copy-pasting can reduce the risk of unknowingly running malicious payloads hidden in copied text.
• Secure your devices. Use an up-to-date, real-time anti-malware solution with a web protection component.
• Learn to recognize evolving attack techniques. Knowing that attackers continually change their methods can help you recognize suspicious instructions. Keep reading our blog!

Pro tip: Did you know that the free Malwarebytes Browser Guard extension warns you when a website tries to copy something to your clipboard?

Some browsers also restrict or warn about certain uses of the clipboard. Since macOS Tahoe 26.4, Terminal can warn users when they paste text copied from a browser or messaging app. This protection operates at the app and operating-system level rather than inside the browser.

---

Stop threats before they can do any harm.

Malwarebytes Browser Guard blocks phishing pages and malicious sites automatically. Free, one click to install. Add it to your browser →

Anthropic has warned some Claude users that criminals are using information stealers to take over their accounts.

Rather than guessing passwords or intercepting two-factor authentication (2FA) codes, the attackers steal the browser sessions that prove a user is already logged in.

According to a warning email shared publicly by an affected user, the attackers used common infostealer malware to copy Claude login sessions from victims’ computers. They then used those sessions to access the accounts and consume their usage.

“We recently signed you out of Claude and removed the payment method saved on your account, so you’ll need to log back in and re-add your card. We’re sorry for the disruption. Here’s what happened and what we’ve done about it.

What happened

We have recently become aware of a bad actor that is using common infostealer malware to steal Claude login sessions from people’s computers, then using those login sessions to access Claude accounts and consume their usage. Our systems detected this activity on your account, and we’ve therefore removed your card on file and signed out the sessions involved to help block further unauthorized access.

If your usage limits looked like they refilled and then drained while you weren’t using Claude, this was likely the cause.”

The message adds that Anthropic has no reason to believe the malware was “related to Claude, installed through Claude, or related to anything you did with Claude.”  

To sum this up:

• Cybercriminals are spreading infostealers. How they are doing this and whether they are targeting groups likely to use Claude professionally is unknown.
• Infostealers can bypass standard credentials and multi-factor authentication (MFA) by stealing active browser sessions and session cookies.
• Once they are able to take over a Claude account, they can consume the victim’s usage and potentially incur additional charges.
• Anthropic is signing affected users out of Claude, removing saved payment methods, and refunding charges it identifies as unauthorized.

To better understand this, you should know that paid Claude plans can offer additional “Usage credits.” When a subscriber reaches the plan’s session limit, Claude can allow them to continue using the service through consumption-based billing at standard API rates. The user must enable the feature, configure a monthly spending limit or select unlimited spending, and prepay for credits.

Users can also enable auto-reload, which automatically buys more prepaid credits when the balance falls below a threshold. So, in a session-hijacking scenario, a thief could use up the account’s included allowance and any available Usage credits. If auto-reload is enabled, they could also trigger further purchases.

The criminals’ likely motive is to use paid Claude capacity for free. The account and any exposed data could also be useful for fraud, social engineering, or follow-on attacks.

Stolen Claude capacity could be used to write and refine phishing and scam content, build campaign infrastructure, develop, modify, or obfuscate malware, improve delivery methods, and analyze stolen information. Cybercriminals can use AI to support several parts of an operation, although Claude has safeguards and abuse monitoring, and Anthropic says it has disrupted accounts used for malicious activity.

What to do

Anthropic provided advice for dealing with a possible infostealer infection. After removing the malware, we recommend you install an up-to-date, real-time anti-malware solution to help protect you against new infections.

These steps are good practice when cleaning up after infostealer malware:

1. Scan any computer you use with Claude for malware and remove any malware before logging back in or changing passwords.
2. Once the malware has been removed, secure the email account you use for Claude by changing its password, signing out of other devices, and enabling two-factor authentication (2FA).
3. Change sensitive passwords that were saved in the affected browser, including those for banking, work, and cloud services. Check your card statements if you stored payment details in the browser.
4. Only after completing these steps should you add your payment method to Claude again if you want your plan to continue renewing.

If you still see your usage changing while Claude is idle, or notice an unrecognized charge after completing these steps, contact usersafety@anthropic.com.

---

From reporting threats to removing them.

Cybersecurity risks should never spread beyond a headline. Keep threats off your devices by downloading Malwarebytes today.

Healthcare and pharmaceutical-distribution giant McKesson has disclosed a cybersecurity incident involving unauthorized access to third-party applications and the theft of data.

McKesson Corporation is an American healthcare company that distributes pharmaceuticals and provides medical supplies, health information technology, and care management tools.

McKesson says it discovered the cybersecurity incident on August 25, 2026, and that its investigation is still in early stages.

“Based on our investigation thus far, including assessments by leading cybersecurity industry experts supporting our response, we’ve confirmed that the unauthorized access to certain third-party applications and the exfiltration of certain data was associated with a subset of customers within our Oncology & Multispecialty and Medical-Surgical business units.”

For now, McKesson provides no information about the amount or nature of the stolen data.

The attack has been claimed by ransomware/extortion group Shiny Hunters. On their leak site the group claims to have stolen hundreds of millions of records containing very sensitive information spanning from Personally Identifiable Information (PII) to Protected Health Information (PHI).

The ShinyHunters extortion group told BleepingComputer that it was behind the attack, claiming it gained access after conducting voice phishing (or vishing) attacks—a form of social engineering—against multiple McKesson employees. Subsequently, the group said it used compromised Okta single-sign-on accounts to access Salesforce and Snowflake environments. It further claimed to have removed approximately 1 TB of data between August 21 and 25.

The group also said the data includes roughly 284 million records, which does not necessarily mean they belong to 284 million unique patients.

A combination of identity information and healthcare-related details could make affected people targets for convincing scams. Criminals could impersonate a pharmacy, insurer, medical provider, debt collector, or patient-support service and use personal details to make the approach appear legitimate.

Healthcare data is especially useful in social-engineering attacks because it can be used to create a sense of urgency: Criminals could scare a target by sending a supposed prescription problem, unpaid claim, delivery issue, appointment change, or request to “verify” insurance details. At this stage, however, McKesson has not confirmed that any particular category of patient data was accessed.

What to do if you’re affected

While waiting for more information about the nature of the breach and how you might be affected, there are a few things you can do:

• Check the company’s advice. Every breach is different, so check with the company to find out what’s happened and follow any specific advice it offers.
• Change your password. You can make a stolen password useless to thieves by changing it. Choose a strong password that you don’t use for anything else. Better yet, let a password manager choose one for you.
• Enable two-factor authentication (2FA). If you can, use a FIDO2-compliant hardware key, laptop, or phone as your second factor. Some forms of 2FA can be phished just as easily as a password, but 2FA that relies on a FIDO2 device can’t be phished.
• Watch out for impersonators. Cybercriminals may contact you posing as the breached company. Check its official website to see if it’s contacting victims, and verify the identity of anyone who contacts you using a different communication channel.
• Take your time. Phishing attacks often impersonate people or brands you know and use themes that require urgent attention, such as missed deliveries, account suspensions, and security alerts.
• Consider not storing your card details. It’s definitely more convenient to let sites remember your card details, but it increases the risk if a company suffers a breach.
• Set up identity monitoring. This can alert you if your personal information is found being traded illegally online and help you recover afterward.

---

Let’s face it, an incognito window can only do so much. 
 
Breaches, dark web trading, credit fraud. Malwarebytes Identity Theft Protection monitors for all of it, alerts you fast, and comes with identity theft insurance. 

Last week on Malwarebytes Labs:

• Protect your WhatsApp account with new passkey and 2FA upgrades
• The AI agent swarm that attacked Hugging Face is a warning for the future
• Flock wants privacy to meet surveillance halfway
• Fake listings can turn trusted platforms into scam springboards
• Fake Apple Pay charge brings the classic tech support scam to your phone
• New Instagram and Facebook rules set a default two-hour limit for teens
• Update Chrome before you browse again
• Popular school apps may be sharing student data with advertisers
• Beware of fake Indeed interview apps used to install spyware
• Grok fooled into stealing user chat, location data, and more
• GTA 6 leak hunt could expose data belonging to thousands of Discord users
• TikTok phishing: How to spot fake login and verification pages
• Fake GTA 6 Extended Look and demo sites deliver an infostealer
• Fake Microsoft security scans trick victims into uninstalling their antivirus
• What happens to your data when you die? (Lock and Code S07E17)
• AliExpress caught using silent audio to fingerprint visitors’ browsers
• ToxicPanda 2.0 can take over your Android phone and banking apps
• Tracking PavinLoader across ClickFix and fake download campaigns

Stay safe!

---

Scammers know more about you than you think. 

Malwarebytes Mobile Security protects you from phishing, scam texts, malicious sites, and more. With real-time AI-powered Scam Guard built right in. 

Download for iOS → Download for Android → 

WhatsApp announced on August 25 that more than one billion people now use passkeys to log back into the app.

The announcement included two other security upgrades: a stronger two-step verification method and more context for incoming calls from unknown numbers. It marks one of the largest passwordless authentication rollouts to date. Passkeys are now firmly mainstream, with the FIDO Alliance estimating that 5 billion are in use worldwide and 75% of consumers have enabled one on at least one account.

Three things changed:

1. Passkey support originally launched on Android and later extended to iOS. WhatsApp now supports multiple passkeys per account, so people who switch between an Android phone and an iPhone (or use both) can register a passkey on each device.
2. Two-step verification is moving from a simple six-digit PIN to a longer alphanumeric password that can include special characters, making it much harder to guess or brute-force.
3. On Android, WhatsApp now shows extra context about calls from numbers not saved in your contacts, including whether the number is from another country and whether you share any groups. It’s a small but useful nudge against the urgency tactics scammers rely on.

Passkeys are resistant to phishing because there is no password or SMS code to type into a fake website or hand over to a scammer. Instead, a passkey is stored on your device or in its credential manager and unlocked using your fingerprint, face, or screen-lock code. They’re also useful in regions where SMS one-time-passcode delivery is unreliable, which might explain why adoption reached a billion users so quickly.

The upgraded two-step verification password closes a real gap. PINs such as “123456” were common, weak, and reused, and a longer alphanumeric password with special characters raises the bar against account-takeover attempts, even if an attacker somehow obtains your one-time code.

The caller-context feature gives people more information to assess legitimacy before answering an unfamiliar number.

---

---

What WhatsApp users need to do

Users need to set up a passkey and upgrade their two-step verification password, while the caller-context feature will appear automatically on supported Android devices:

• Set up a passkey via Settings > Account > Passkeys, and follow the instructions on your device. Don’t forget to add a second one if you use both an Android and an iOS device.
• If you still use a six-digit PIN for two-step verification, upgrade to the new password format when it becomes available, especially if your PIN is predictable. You can find instructions to set up two-step verification for WhatsApp in this blog. If it’s already enabled, select Two-step verification to find the option to change your PIN.
• Add a recovery email to two-step verification if you haven’t already. It’s the only way to reset the password if you forget it.
• Android users should pay attention to the new caller-context details before answering calls from unknown numbers, treating urgency as a red flag rather than a reason to rush.

Passkeys and stronger two-step verification aren’t retroactive or forced, so accounts still relying on an old PIN or no passkey at all will remain unchanged until users upgrade them.

---

Scammers know more about you than you think. 

Malwarebytes Mobile Security protects you from phishing, scam texts, malicious sites, and more. With real-time AI-powered Scam Guard built right in. 

Download for iOS → Download for Android → 

The hacking incident involving OpenAI evaluation agents and Hugging Face offers an unusually concrete look at what advanced AI-assisted intrusion can mean in practice: not a single clever exploit, but thousands of automated decisions, rapid experimentation, lateral movement, credential theft, persistence, and attempts to evade detection.

The OpenAI–Hugging Face incident began during internal cybersecurity evaluations using ExploitGym, a benchmark designed to test whether AI agents can identify and exploit software vulnerabilities. OpenAI says the affected agents were meant to operate in isolated cloud sandboxes, with no internet access or communication between agents in many of the evaluation environments.

How the agents got around those boundaries

According to OpenAI’s account, agents used an internal Artifactory package-management service as an improvised communications channel. By writing files and directory names that other agents could read, isolated instances effectively created their own message board.

This meant the agents were no longer acting merely as independent benchmark runners. They could share discoveries, leave instructions, divide work, and build on one another’s successes.

An independent investigation by METR and Redwood Research found that roughly 1,200 agents used the unauthorized message board, exchanging more than 70,000 messages and files. About 700 of them went on to participate in the attack on Hugging Face.

The scale is the threat

A model capable of finding one useful exploit presents a risk. Hundreds of agents able to pool information, delegate tasks, and repeatedly test routes around a control boundary become a much more difficult problem.

Hugging Face reconstructed about 17,600 attacker actions across roughly four and a half days, including reconnaissance, remote code execution, credential access, command-and-control activity, Kubernetes enumeration, movement through the internal network, and supply-chain probing.

A human intruder must choose where to spend time. An agent swarm can continuously test ideas, retry failed paths, switch infrastructure when blocked, preserve notes for other agents, and run many experiments in parallel. It can also create noise: Thousands of low-confidence actions, most of which lead nowhere, can obscure the handful that matter.

Hugging Face described this as an “asymmetry problem.” An attacker can generate enormous volumes of attempts, while defenders must determine which signals represent a real escalation before the attacker reaches a sensitive system.

What needs to change

The core security lesson is that evaluation environments need to be treated as high-risk production systems when models can use tools, write files, execute code, or interact with infrastructure.

The technology is rapidly becoming capable of doing more. The question is whether the people building, deploying, and defending it are ready for what comes next.

OpenAI says it has since moved toward stronger workload and network isolation, tighter control over internet access and model weights, and mandatory chain-of-thought monitoring for high-capability, tool-using training and evaluations.

“We are taking this incident as a ‘warning shot’ that today’s model capabilities present the possibility of loss-of-control incidents. Our security and alignment posture is escalating accordingly. These events also highlight risks in future AI development that extend beyond OpenAI and will require the attention of the whole industry.”

At the same time, a US court ruling involving Anthropic has brought a related question into focus: whether developers can place limits on high-risk uses of their models, including mass surveillance and fully autonomous weapons.

The legal dispute is political by nature, but its technical underpinning is hard to ignore. If capable AI systems can enhance offensive cyberattack methods and bypass safety restrictions, access controls, logging, and deployment boundaries, those safeguards are no longer abstract policy choices.

Advanced AI agents can be useful to defenders as well as attackers. But the surrounding systems need to be trusted to keep their capabilities bounded when something goes wrong.

Who benefits from more capable AI?

The security debate around AI agents often focuses on whether systems can be controlled. Can they be kept inside a sandbox? Can their tools, credentials, network access, and autonomy be restricted? Can defenders detect harmful behavior before it becomes an incident?

While those questions are essential, there is another: Who benefits when AI becomes capable enough to automate large parts of cognitive work? Who carries the costs when it fails, displaces workers, enables fraud, causes damage, or concentrates power?

AI could give small organizations access to technical expertise that previously required large teams and budgets. It could help doctors identify urgent cases sooner, help teachers tailor support to individual students, assist people with disabilities, speed up scientific research, and make complex public services easier to navigate. For cybersecurity teams, it could make vulnerability triage, alert investigation, threat hunting, and incident response faster and more accessible.

Bill Gates has argued that while AI could bring remarkable benefits to health care, education, agriculture, scientific research, and public services, the outcome will depend on deliberate choices rather than technical progress alone. He also warns that AI’s rapid adoption could widen inequality, disrupt entry-level and mid-career work, make harmful capabilities more accessible, and reinforce existing concentrations of power.

Gates also argues that “self-regulation on the most dangerous tool ever invented” does not sound like a good idea.

“AI will either be the greatest equalizer ever invented, or the worst source of injustice.”

Right now, we still have a choice.

---

Let’s face it, an incognito window can only do so much. 
 
Breaches, dark web trading, credit fraud. Malwarebytes Identity Theft Protection monitors for all of it, alerts you fast, and comes with identity theft insurance. 

Flock Safety CEO Garrett Langley says the United States needs a “compromise” between privacy and public safety.

It’s a neat phrase, except I don’t like to see “compromise” and “privacy” that close together.

“When people talk about just one of these, privacy or safety, they’re prioritizing the wrong thing, and what we have to prioritize as a country is compromise.”

Langley call for compromise comes as the company faces intensifying resistance to its automated license plate reader (ALPR) network. The opposition has begun to affect Flock commercially and operationally, with agencies disabling cameras or canceling contracts.

The problem is that the public has already been doing the compromising: People’s movements have been routinely captured, stored, searched, and, in some cases, shared far beyond the communities that installed the cameras.

Flock’s ALPRs collect detailed records of where vehicles travel, then make that data available to law enforcement for investigations. Langley now says the company wants more regulation and accountability. Flock says it’s reducing its recommended default retention period to seven days and will require case codes for law enforcement and an audit tool designed to flag suspicious access by the end of the year.

Those are welcome concessions, but they do not resolve the underlying concern: A rapidly expanding, privately operated surveillance network can turn ordinary travel into searchable historical data.

The opposition has not faded; it has intensified. NPR reports that cameras have been vandalized in at least 36 states. Vandalism is neither a productive nor lawful answer, but its spread offers a useful measure of how profoundly many people feel excluded from decisions about surveillance in their communities.

A genuine compromise would not start with the assumption that widespread collection is inevitable and then negotiate the retention period. It would begin with democratic consent, strict limits on how the data can be used, independently enforceable access controls, public reporting, meaningful opt-outs where possible, and a clear requirement that surveillance be necessary and proportionate.

Calls to meet halfway are also harder to take seriously when the CEO has been accused by 404 Media of misleading police about the outlet’s reporting on an abortion-related case. According to 404 Media, his account is contradicted by court records and police reports. That accusation makes his public calls for compromise much harder to accept.

Flock is right about one thing: There needs to be accountability. But calling for “compromise” after the cameras are already up sharply limits the choices left to communities. Privacy is not a bargaining chip to be surrendered whenever surveillance vendors promise safety.

---

Browse like no one’s watching. 

Malwarebytes Privacy VPN encrypts your connection and never logs what you do, so the next story you read doesn’t have to feel personal. Try it free →