Today’s Topics:

• 7-Zip Symbolic Link Flaw Draws Attention After Public PoC Release
• Another Salesforce Supply-Chain Breach: Gainsight Compromise Fuels OAuth Token Theft
• How can Netizen help?

7-Zip Symbolic Link Flaw Draws Attention After Public PoC Release

Reports from NHS England Digital briefly suggested that a newly disclosed flaw in 7-Zip was being used in real attacks, but the agency later corrected its advisory, clarifying that it has not seen evidence of live exploitation. What they have confirmed is the presence of a public proof-of-concept, which raises the stakes for anyone still running outdated versions of the tool.

The issue, tracked as CVE-2025-11001, affects how 7-Zip processes symbolic links inside ZIP archives. A crafted archive can push the program into unintended directories and open the door for remote code execution under a service-level account. Trend Micro’s ZDI highlighted the directory traversal weakness last month, and the fix quietly arrived with version 25.00 in July. The flaw was introduced several versions earlier, making long-term installs especially exposed.

Researchers Ryota Shiga and Takumi, an AI-assisted auditing system from GMO Flatt Security, discovered and disclosed the problem. A second, similar bug, CVE-2025-11002, was also fixed in the same release and involves the same symbolic-link handling weakness. Both issues share the same severity score and the same potential impact.

Although NHS initially suggested active exploitation, the updated advisory walks that back and attributes the earlier wording to an error. What remains true is that a PoC is already available. Security researcher Dominik, who published the demonstration, noted that successful exploitation requires either an elevated account or Windows developer mode. The vulnerability only affects Windows systems and cannot trigger outside those conditions.

With public exploit material already circulating, users relying on older 7-Zip versions are exposed to unnecessary risk. Updating to version 25.00 or later closes both symbolic-link flaws and prevents attackers from using crafted archives to gain footholds on a target system.

Another Salesforce Supply-Chain Breach: Gainsight Compromise Fuels OAuth Token Theft

Salesforce customers are once again dealing with a familiar and avoidable problem: attackers abusing third-party integrations to slip into environments that organizations assumed were already under control. The newest incident mirrors the Drift breach from earlier in the year, only this time the attackers used Gainsight as their entry point. OAuth tokens tied to Gainsight’s connection with Salesforce were stolen, giving the threat group access to customer environments with whatever permissions each organization had granted the app.

The attackers behind this campaign are linked to the ShinyHunters extortion group, which has spent much of the past year targeting SaaS integrations that provide broad access but are often treated as low-risk. Google’s threat intelligence team attributed this latest wave to a group connected to ShinyHunters and said that more than 200 Salesforce environments were affected. The attackers themselves claimed nearly 1,000 across both Drift and Gainsight.

Salesforce responded by pulling the affected apps from its marketplace and revoking all active OAuth tokens associated with Gainsight. That decision briefly caused confusion inside Gainsight, which initially believed the sudden failure of customer connections was a technical glitch. Salesforce later clarified that revoking the tokens did not erase audit trails or limit customers’ ability to investigate the breach.

The most striking part of this episode is how straightforward the attackers’ strategy was. Security researchers pointed out that Drift never required the level of access many customers had given it, and the same pattern repeated with Gainsight. These integrations were granted extensive permissions far beyond what a sales-oriented tool reasonably needs, creating a perfect opportunity for attackers once those OAuth tokens were stolen.

This isn’t just a Salesforce issue. Gainsight connects to a long list of other platforms; Slack, Microsoft Teams, HubSpot, Jira, Snowflake, and many more. Any organization that integrated Gainsight without a clear access policy may now be exposed across several systems, not just Salesforce. Many teams are only now realizing how many places their SaaS tools connect and how little visibility they actually have.

How Can Netizen Help?

Founded in 2013, Netizen is an award-winning technology firm that develops and leverages cutting-edge solutions to create a more secure, integrated, and automated digital environment for government, defense, and commercial clients worldwide. Our innovative solutions transform complex cybersecurity and technology challenges into strategic advantages by delivering mission-critical capabilities that safeguard and optimize clients’ digital infrastructure. One example of this is our popular “CISO-as-a-Service” offering that enables organizations of any size to access executive level cybersecurity expertise at a fraction of the cost of hiring internally. 

Netizen also operates a state-of-the-art 24x7x365 Security Operations Center (SOC) that delivers comprehensive cybersecurity monitoring solutions for defense, government, and commercial clients. Our service portfolio includes cybersecurity assessments and advisory, hosted SIEM and EDR/XDR solutions, software assurance, penetration testing, cybersecurity engineering, and compliance audit support. We specialize in serving organizations that operate within some of the world’s most highly sensitive and tightly regulated environments where unwavering security, strict compliance, technical excellence, and operational maturity are non-negotiable requirements. Our proven track record in these domains positions us as the premier trusted partner for organizations where technology reliability and security cannot be compromised.

Netizen holds ISO 27001, ISO 9001, ISO 20000-1, and CMMI Level III SVC registrations demonstrating the maturity of our operations. We are a proud Service-Disabled Veteran-Owned Small Business (SDVOSB) certified by U.S. Small Business Administration (SBA) that has been named multiple times to the Inc. 5000 and Vet 100 lists of the most successful and fastest-growing private companies in the nation. Netizen has also been named a national “Best Workplace” by Inc. Magazine, a multiple awardee of the U.S. Department of Labor HIRE Vets Platinum Medallion for veteran hiring and retention, the Lehigh Valley Business of the Year and Veteran-Owned Business of the Year, and the recipient of dozens of other awards and accolades for innovation, community support, working environment, and growth.

Looking for expert guidance to secure, automate, and streamline your IT infrastructure and operations? Start the conversation today.

Today’s Topics:

• The First AI-Driven Cyber Espionage Campaign Signals a Turning Point for Global Security
• Zero-Day Exploits in Cisco ISE and Citrix NetScaler Show Attackers’ Shift Toward Identity Infrastructure
• How can Netizen help?

The First AI-Driven Cyber Espionage Campaign Signals a Turning Point for Global Security

In September 2025, researchers uncovered what appears to be the first large-scale cyber espionage operation driven almost entirely by artificial intelligence. A Chinese state-sponsored group reportedly used Anthropic’s Claude Code model to infiltrate around thirty major organizations, including technology firms, financial institutions, manufacturers, and government agencies. The attack’s defining feature was its autonomy. Instead of relying on coordinated human operators, the attackers built an AI-based framework that executed reconnaissance, vulnerability testing, and data exfiltration on its own.

The campaign began with a jailbreak that tricked Claude Code into believing it was performing legitimate security testing for a cybersecurity firm. Once the system was compromised, the attackers assigned it a series of small, context-limited tasks to bypass the model’s guardrails. From there, the AI analyzed target networks, located sensitive databases, and wrote exploit code without human instruction. It harvested credentials, identified privileged accounts, and exported large quantities of data. It even created its own documentation, cataloging stolen credentials and mapping system structures for future reference. Roughly eighty to ninety percent of the entire campaign was conducted autonomously, with humans stepping in only a handful of times to provide direction or review results.

Until now, even the most advanced campaigns required continuous human oversight. This one demonstrated that AI can perform sustained attacks across multiple organizations at speeds that no human team could match. The model issued thousands of requests every second, performing reconnaissance and exploitation simultaneously across several targets. Although it occasionally produced false data or exaggerated its success, its accuracy was still high enough to compromise real systems and extract genuine credentials.

Anthropic’s internal teams acted quickly once the activity was detected. They disabled malicious accounts, informed affected entities, and worked with international authorities to analyze the full scope of the campaign. They have since improved their classifiers for detecting malicious use and developed stronger monitoring systems to identify AI-generated attack traffic. Despite these steps, the broader risk remains that similar methods could be used on other large language models or autonomous agents that integrate with external tools.

Zero-Day Exploits in Cisco ISE and Citrix NetScaler Show Attackers’ Shift Toward Identity Infrastructure

In May 2025, Amazon’s threat intelligence division identified a coordinated intrusion campaign that exploited two previously unknown zero-day vulnerabilities in Cisco Identity Services Engine (ISE) and Citrix NetScaler ADC. The operation delivered a custom-built web shell through precision-crafted payloads, reflecting a growing trend of attackers focusing on identity and access infrastructure that manages authentication and policy enforcement across enterprise networks.

Amazon’s MadPot honeypot network was the first to detect abnormal traffic associated with CVE-2025-5777, later dubbed Citrix Bleed 2. The flaw, a critical input validation error in Citrix NetScaler ADC and Gateway, carried a CVSS score of 9.3 and allowed attackers to bypass authentication entirely. Citrix addressed the issue with a patch in June 2025, but the exploit had already been active weeks earlier. During the investigation, Amazon uncovered a second vulnerability being exploited in parallel. CVE-2025-20337, found in Cisco ISE and its Passive Identity Connector, allowed unauthenticated remote code execution as root and received the highest possible CVSS rating of 10.0. Cisco issued a patch for it in July 2025.

The threat actor used these vulnerabilities together to install a stealthy backdoor disguised as a legitimate Cisco ISE module named IdentityAuditAction. Unlike common off-the-shelf malware, this implant was written specifically for ISE environments. It operated entirely in memory, using Java reflection to inject itself into existing Tomcat threads. Once established, it registered as an HTTP listener that monitored inbound requests while encrypting data with DES and encoding it in a modified Base64 format. These techniques helped it remain undetected within normal Cisco operations and maintain persistence without writing files to disk.

Amazon assessed the group behind the campaign as highly skilled and well-funded. The use of two zero-days at once indicated access to either internal vulnerability research or privileged information unavailable to the public. The attackers demonstrated deep familiarity with enterprise Java systems and Cisco ISE internals, suggesting that the campaign had been carefully planned and executed over an extended period.

Although the targeting appeared broad rather than aimed at specific victims, the implications are serious. Network appliances like Citrix NetScaler and Cisco ISE control authentication, segmentation, and access policy enforcement. A compromise of these systems provides adversaries with extensive control over network trust boundaries, potentially neutralizing key elements of Zero Trust security frameworks.

CJ Moses, Chief Information Security Officer for Amazon Integrated Security, noted that the pre-authentication nature of these vulnerabilities means even well-maintained environments can be compromised. He emphasized that organizations must isolate administrative portals behind firewalls, enforce strict access policies, and deploy monitoring tools that can recognize abnormal web server behavior indicative of exploitation.

These findings reinforce the importance of adopting a defense-in-depth approach. Network access control systems must be treated with the same scrutiny as endpoints and cloud services. Continuous exposure testing, patch verification, and behavioral analytics can help identify anomalies before attackers exploit them.

How Can Netizen Help?

Founded in 2013, Netizen is an award-winning technology firm that develops and leverages cutting-edge solutions to create a more secure, integrated, and automated digital environment for government, defense, and commercial clients worldwide. Our innovative solutions transform complex cybersecurity and technology challenges into strategic advantages by delivering mission-critical capabilities that safeguard and optimize clients’ digital infrastructure. One example of this is our popular “CISO-as-a-Service” offering that enables organizations of any size to access executive level cybersecurity expertise at a fraction of the cost of hiring internally. 

Netizen also operates a state-of-the-art 24x7x365 Security Operations Center (SOC) that delivers comprehensive cybersecurity monitoring solutions for defense, government, and commercial clients. Our service portfolio includes cybersecurity assessments and advisory, hosted SIEM and EDR/XDR solutions, software assurance, penetration testing, cybersecurity engineering, and compliance audit support. We specialize in serving organizations that operate within some of the world’s most highly sensitive and tightly regulated environments where unwavering security, strict compliance, technical excellence, and operational maturity are non-negotiable requirements. Our proven track record in these domains positions us as the premier trusted partner for organizations where technology reliability and security cannot be compromised.

Netizen holds ISO 27001, ISO 9001, ISO 20000-1, and CMMI Level III SVC registrations demonstrating the maturity of our operations. We are a proud Service-Disabled Veteran-Owned Small Business (SDVOSB) certified by U.S. Small Business Administration (SBA) that has been named multiple times to the Inc. 5000 and Vet 100 lists of the most successful and fastest-growing private companies in the nation. Netizen has also been named a national “Best Workplace” by Inc. Magazine, a multiple awardee of the U.S. Department of Labor HIRE Vets Platinum Medallion for veteran hiring and retention, the Lehigh Valley Business of the Year and Veteran-Owned Business of the Year, and the recipient of dozens of other awards and accolades for innovation, community support, working environment, and growth.

Looking for expert guidance to secure, automate, and streamline your IT infrastructure and operations? Start the conversation today.

Today’s Topics:

• ClickFix Phishing Wave Hits Hotels and Hijacks Booking Accounts With PureRAT
• Microsoft Warns of Whisper Leak: Encrypted AI Chat Traffic Can Reveal User Topics
• How can Netizen help?

ClickFix Phishing Wave Hits Hotels and Hijacks Booking Accounts With PureRAT

Large-scale phishing activity is hitting the hospitality sector again, and researchers say the latest wave is using convincing ClickFix-style pages to push PureRAT onto hotel systems. The operation has been active since spring 2025 and appears to have accelerated through early fall, with attackers focusing on hotel managers who maintain Booking.com, Expedia, and other reservation platforms.

The attack starts with email accounts that have already been compromised. From there, hotel staff receive messages that look like legitimate booking updates or verification prompts. When they click through, they’re sent to a fake verification page that imitates a reCAPTCHA step. That page then urges them to run a copied command on their computer. Once executed, the command retrieves a ZIP archive containing a binary that uses DLL side-loading to load PureRAT.

PureRAT gives the attacker broad control. It can log keystrokes, capture webcam and microphone feeds, move files in and out, proxy traffic, run commands, and maintain persistent access through a Run registry key. The malware is packed with protections that complicate reverse engineering, making analysis slower and giving the operators more time with a compromised system.

Once threat actors gain access to hotel extranet accounts, they use or sell the stolen credentials. These accounts are valuable because they allow direct contact with guests. Attackers send messages over email or WhatsApp containing accurate reservation information, then guide customers to fake landing pages that imitate Booking.com or Expedia. The goal is to collect card details under the false pretext of preventing cancellations or verifying payment.

Behind the scenes, the scheme relies heavily on underground marketplaces. Criminal groups buy and sell Booking.com, Expedia, Airbnb, and Agoda logs, often bundled as username-password pairs or session cookies harvested from infostealer infections. Log-checker tools and Telegram bots make it easy for buyers to validate that the stolen accounts still work, which keeps the cycle running smoothly.

The sophistication of the ClickFix technique continues to grow. Newer versions of the phishing page display a short countdown timer, a fake verification counter, and even embedded videos to make the prompt feel routine and harmless. The page adapts to the victim’s operating system, giving system-specific instructions and automatically copying the malicious command to the clipboard to reduce friction.

This is part of a broader trend: fraud groups are building repeatable, service-based workflows around these attacks. Compromise leads to credential harvesting, which leads to guest-targeting scams, all supported by cheap tools, malware distributors, and credential brokers. As these pages become more convincing, hotel staff and customers become easier targets.

Microsoft Warns of Whisper Leak: Encrypted AI Chat Traffic Can Reveal User Topics

Microsoft is warning about a new privacy threat called Whisper Leak, a side-channel technique that allows someone watching encrypted traffic to guess what topics a user is discussing with an AI chatbot. Even though the traffic is protected with TLS, packet sizes and timing patterns still reveal enough structure for an attacker with the right access to narrow down conversation themes.

According to Microsoft’s researchers, an attacker positioned at an ISP, on a shared network, or on the same Wi-Fi could collect encrypted packets, analyze their sequence, and use machine learning models to classify whether the user’s prompt matches a topic of interest. This works because streaming models send data incrementally, and those streams often reflect token boundaries and response pacing in ways that can be measured even without decrypting the content.

Microsoft’s tests used LightGBM, Bi-LSTM, and BERT classifiers to determine whether a prompt belonged to a specific target category. Several prominent models from major vendors were found to be vulnerable, with classification rates above 98 percent in many cases. Google and Amazon models showed more resistance, likely due to their token batching methods, though they were not completely unaffected.

This raises clear concerns. If a surveillance actor collected enough traffic over time, they could reliably flag users asking about sensitive subjects, whether political, financial, or otherwise monitored. The technique also becomes stronger as the attacker gathers more samples to train on, making long-term monitoring more effective than one-off observations.

Vendors have started deploying mitigations. The most effective countermeasure adds a random, variable-length text segment to each streamed output, which disrupts the relationship between token size and packet size. Microsoft, OpenAI, Mistral, and xAI have already incorporated these defenses.

In the meantime, users who are concerned about privacy are advised to avoid discussing sensitive topics on insecure networks, use VPNs, or rely on non-streaming model modes. Choosing providers that have implemented Whisper Leak countermeasures can also limit exposure.

This disclosure arrives alongside another study showing that many open-weight models remain vulnerable to multi-turn adversarial prompts. Researchers found that safety degradation becomes more pronounced across longer conversations, especially in models designed primarily for capability instead of safety. These findings reinforce that organizations deploying open-source or lightly-aligned models still face meaningful risks unless they apply additional security controls, perform regular red-team testing, and maintain strict system-prompt guidance.

How Can Netizen Help?

Founded in 2013, Netizen is an award-winning technology firm that develops and leverages cutting-edge solutions to create a more secure, integrated, and automated digital environment for government, defense, and commercial clients worldwide. Our innovative solutions transform complex cybersecurity and technology challenges into strategic advantages by delivering mission-critical capabilities that safeguard and optimize clients’ digital infrastructure. One example of this is our popular “CISO-as-a-Service” offering that enables organizations of any size to access executive level cybersecurity expertise at a fraction of the cost of hiring internally. 

Netizen also operates a state-of-the-art 24x7x365 Security Operations Center (SOC) that delivers comprehensive cybersecurity monitoring solutions for defense, government, and commercial clients. Our service portfolio includes cybersecurity assessments and advisory, hosted SIEM and EDR/XDR solutions, software assurance, penetration testing, cybersecurity engineering, and compliance audit support. We specialize in serving organizations that operate within some of the world’s most highly sensitive and tightly regulated environments where unwavering security, strict compliance, technical excellence, and operational maturity are non-negotiable requirements. Our proven track record in these domains positions us as the premier trusted partner for organizations where technology reliability and security cannot be compromised.

Netizen holds ISO 27001, ISO 9001, ISO 20000-1, and CMMI Level III SVC registrations demonstrating the maturity of our operations. We are a proud Service-Disabled Veteran-Owned Small Business (SDVOSB) certified by U.S. Small Business Administration (SBA) that has been named multiple times to the Inc. 5000 and Vet 100 lists of the most successful and fastest-growing private companies in the nation. Netizen has also been named a national “Best Workplace” by Inc. Magazine, a multiple awardee of the U.S. Department of Labor HIRE Vets Platinum Medallion for veteran hiring and retention, the Lehigh Valley Business of the Year and Veteran-Owned Business of the Year, and the recipient of dozens of other awards and accolades for innovation, community support, working environment, and growth.

Looking for expert guidance to secure, automate, and streamline your IT infrastructure and operations? Start the conversation today.