Cyber Alerts

Two separate incidents involving Amazon and Google show that AI-powered development tools are now viable targets for attackers. With deep access to developer environments, weak authentication controls, and high trust, these tools create new risk surfaces across the software supply chain. Incident #1: Amazon Q Developer Extension A malicious actor injected code into Amazon Q’s GitHub repo via an unverified pull request. The altered version (1.84.0) was published to nearly a million users through the VS Code marketplace. While the code was malformed, it demonstrated how workflow misconfigurations can turn helpful extensions into attack vectors. Incident #2: Google Gemini CLI Flaw A vulnerability in Gemini CLI’s setup workflow allowed attackers to hijack authentication tokens and silently execute commands on dev systems. The issue was fixed, but not before showing how token handling flaws can create remote execution paths. Recommended Actions: - Patch Amazon Q Developer to v1.85.0 - Audit permission scopes for CLI and IDE extensions - Harden GitHub workflows and authentication tokens - Treat AI development tools as privileged software—because they are

CISA has confirmed active exploitation of CVE‑2023‑2533, a remote code execution vulnerability in PaperCut NG/MF. The flaw allows unauthenticated attackers to execute code on vulnerable servers via the SetupCompleted configuration page. What You Should Know: - Exploited in the wild, this bug poses a major risk to organizations using PaperCut to manage print services. - - Attackers can bypass authentication, run malicious code, and gain full control of the affected systems. Recommended Actions: - Apply PaperCut patches immediately - Restrict PaperCut access to internal networks only - Review logs for any unusual SetupCompleted access - Isolate print servers from business-critical assets

Scattered Spider, known for targeting retailers, has expanded into insurance and now airline sectors. Their updated campaign uses deepfake video calls, vishing, and in-platform impersonations to breach MFA and trick employees. What You Should Know: - Google TAG confirms the shift to insurance; the FBI warns airlines are now at risk. - The group leverages human trust—posing as IT support to bypass MFA—and uses collaboration apps and deepfake tools to access systems without technical exploits. Recommended Actions: - Educate staff on multi-channel impersonation across video, voice, and chat - Restrict remote-access tool installations to official channels - Monitor for anomalous activity in help-desk and ticketing platforms - Verify all MFA and support requests with strict multi-factor verification

Citrix has disclosed a new critical vulnerability in NetScaler ADC and Gateway (CVE-2025-5777), dubbed CitrixBleed 2. The flaw allows unauthenticated attackers to extract sensitive data directly from system memory, including session tokens, credentials, and encryption keys. What You Should Know: - Though no exploitation has yet been confirmed, experts predict that attacks are likely imminent. - The vulnerability affects common NetScaler configurations for remote access (VPN, ICA Proxy, CVPN, RDP Proxy) used by many large enterprises. - Attackers can leverage this weakness without user interaction or credentials. Recommended Actions: - Apply Citrix security updates immediately - Verify all external-facing NetScaler appliances are patched - Execute Citrix-recommended commands post-patch to terminate active sessions - Monitor logs for signs of unauthorized session access

Microsoft has patched a zero-click vulnerability in Copilot, the AI assistant integrated into Microsoft 365. The flaw allowed attackers to extract internal data by embedding malicious prompts in emails—without the user clicking or responding. What You Should Know: This is the first known exploit of its kind involving an enterprise AI assistant. The vulnerability—CVE-2025-32711—was significant because it demonstrated how prompt injection could be weaponized for real-world data theft without user interaction. The incident underscores the importance of visibility and control over AI-assisted workflows. Recommended Actions: - Audit Copilot activity logs for unusual data access patterns - Evaluate where AI may interact with sensitive business information - Train users on risks tied to AI-generated email content - Begin establishing monitoring protocols for future AI-driven features

Google’s Threat Analysis Group has warned of a campaign in which attackers are exploiting OAuth permissions to gain persistent access to Salesforce accounts. Once in, they install malicious applications capable of data theft or malware deployment. What You Should Know: OAuth-based attacks bypass login credentials entirely. Traditional security controls—like MFA or login alerts—don’t detect this. The risk is elevated for any organization with widely used or lightly monitored SaaS platforms. Recommended Actions: - Review and restrict third-party app access in Salesforce - Audit OAuth token logs and scopes for abnormalities - Remove unused or suspicious integrations - Train staff on the risks of app authorization and fake consent screens

China-linked APT41 is leveraging Google Calendar to stealthily deliver malware. Victims receive calendar invites with malicious URLs disguised as event details, leading to second-stage malware. What You Should Know: This method blends into daily workflow and is unlikely to trigger traditional email filters. It’s a reminder that even trusted cloud platforms can be weaponized. Recommended Actions - Monitor for abnormal calendar event creation - Enforce calendar URL scanning and restriction policies - Train staff to flag suspicious event invites - Evaluate G Workspace settings for risk reduction

Threat actors have created a convincing fake Bitdefender website to spread malware. Unsuspecting users who download software from the site receive VenomRAT or a credential-harvesting infostealer. What You Should Know: This campaign targets users seeking legitimate antivirus protection—turning their search into an infection. Malicious downloads bypass some endpoint tools via trusted branding. Recommended Actions - Block impersonation domains and scan installer traffic - Educate users on software sourcing hygiene - Monitor for RAT behavior and credential leaks - Review endpoint protection efficacy against sideloaded threats

New Interlock ransomware is being deployed alongside NodeSnake, a JavaScript-based RAT. This tactic allows threat actors to silently exfiltrate data before executing encryption, increasing the impact and complexity of response. What You Should Know: This threat is a dual-pronged attack. Even with backups in place, stolen data can lead to compliance risks, legal exposure, and leverage for extortion. Detection and prevention must extend beyond traditional ransomware defenses. Recommended Actions: - Strengthen detection rules for JavaScript and fileless malware activity. - Monitor endpoints and cloud workloads for NodeSake behavior. - Enforce DLP policies and inspect outbound traffic for exfiltration signs. - Adapt ransomware response plans to cover data theft before encryption.

CISA has issued a new alert detailing active exploitation of critical vulnerabilities in Commvault software—a widely used backup and recovery platform. Threat actors are leveraging these flaws to gain remote access, exfiltrate data, and move laterally into cloud environments. What You Should Know: These vulnerabilities are actively being exploited in the wild—this is not a drill. If left unaddressed, they create a clear path for attackers to escalate privilege and compromise critical cloud systems. Steps to take now: Patch all exposed Commvault instances—especially those accessible via the internet. Audit cloud and backup integrations for unusual behavior. Strengthen identity controls to prevent lateral movement. Hunt for activity tied to CVE-2025-3928 and related indicators of compromise.

A joint advisory from the U.S., Australia, and Canada highlights a growing ransomware tactic: fast-flux DNS — a technique that rapidly rotates IPs and domains to evade detection and takedown. Cybercriminal groups like LockBit and Black Basta, with links to Russia, are actively using it to target sectors including healthcare, government, and critical infrastructure. Why it matters to you: •Fast-flux makes ransomware attacks harder to trace, contain, and mitigate •Traditional defenses may not detect this level of infrastructure agility •The tactic signals a continued evolution in attacker sophistication 📌 Next steps: Ensure your teams are evaluating DNS-layer defense strategies and securing offline backups.

A critical Windows zero-day vulnerability, identified as ZDI-CAN-25373, is currently being actively exploited by nation-state threat actors. This flaw allows attackers to execute hidden malicious commands via specially crafted Windows shortcut (.lnk) files. The vulnerability has been present for at least eight years, was reported six months ago but remains unpatched by Microsoft. Nation-state actors from North Korea, Iran, Russia, and China are leveraging this exploit to conduct cyber espionage and financial crimes, particularly targeting cryptocurrency platforms and sensitive government data. To mitigate the risk, please take the following immediate actions: •Monitor for suspicious .lnk files: Scan systems for shortcut files that may contain embedded malicious commands. •Restrict execution of unknown shortcuts: Prevent the automatic execution of .lnk files from untrusted sources. •Use endpoint detection and response (EDR) solutions: Deploy advanced security tools to detect anomalous behavior linked to shortcut exploitation. •Educate employees: Train staff to recognize suspicious files and avoid executing unknown shortcuts. Given the severity and active exploitation of ZDI-CAN-25373, it is imperative to assess your exposure and strengthen defenses accordingly.

A critical Windows zero-day vulnerability, identified as ZDI-CAN-25373, is currently being actively exploited by nation-state threat actors. This flaw allows attackers to execute hidden malicious commands via specially crafted Windows shortcut (.lnk) files. The vulnerability has been present for at least eight years, was reported six months ago but remains unpatched by Microsoft. Nation-state actors from North Korea, Iran, Russia, and China are leveraging this exploit to conduct cyber espionage and financial crimes, particularly targeting cryptocurrency platforms and sensitive government data. To mitigate the risk, please take the following immediate actions: •Monitor for suspicious .lnk files: Scan systems for shortcut files that may contain embedded malicious commands. •Restrict execution of unknown shortcuts: Prevent the automatic execution of .lnk files from untrusted sources. •Use endpoint detection and response (EDR) solutions: Deploy advanced security tools to detect anomalous behavior linked to shortcut exploitation. •Educate employees: Train staff to recognize suspicious files and avoid executing unknown shortcuts. Given the severity and active exploitation of ZDI-CAN-25373, it is imperative to assess your exposure and strengthen defenses accordingly.