CyberBrief Project

CyberBrief Project

Download on the App Store

CyberBrief Project episodes

  • Inside Chaos: The New Face of Ransomware

    "Send me a quick text"

    Chaos is a new ransomware group making its mark with aggressive campaigns and calculated pressure on victims. What appears to be a fresh name is, in fact, a continuation of familiar strategies, designed to confuse analysts and buy the attackers more time.

    In this episode, we break down how Chaos positions itself in the ransomware landscape, why its approach is so disruptive, and what defenders can learn from the group’s focus on leverage, pressure, and rebranding. The story highlights the broader trend of ransomware operations evolving their identity while keeping proven methods intact.

    Defensive priorities

    • Restrict or gate Quick Assist usage, and require verification for any phone-based IT requests.
    • Monitor for new or unsanctioned installations of AnyDesk, ScreenConnect, Splashtop, or OptiTune.
    • Detect GoodSync activity or executables presenting as wininit.exe that transfer files to external storage.
    • Hunt for SSH over port 443 that uses no known_hosts entries and skips host key verification.
    • Alert on registry changes that hide accounts, removal of PowerShell logs, and shadow copy deletion.
    • Flag files carrying the chaos file extension and ransom notes named readme.chaos.txt.

    Selected IOCs and tools

    • Reverse tunnel host: 45.61.134.36 on port 443
    • Encrypted extension: the chaos file extension
    • Ransom note: readme.chaos.txt
    • Abused tools: Quick Assist, Impacket, GoodSync, AnyDesk, ScreenConnect, Splashtop

    Detection should emphasize unusual RDP, SMB, and WMI activity, signs of Impacket usage, and credential harvesting behaviors consistent with Kerberoasting. Strong MFA enforcement and continuous endpoint monitoring remain essential.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    8 min
  • The Reply Trap: Instagram Phishing Without a Website

    "Send me a quick text"

    A recent phishing campaign is targeting Instagram users with messages that closely resemble legitimate account alerts. Victims are tricked into responding in ways that validate their address and gradually hand over details attackers can use to reset passwords, hijack accounts, and pivot into connected services. Once an account is taken, it can be locked for ransom, abused to spread further scams, or sold on underground markets.

    Key defensive actions:

    • Train users to verify the sender and question unexpected account-security messages.
    • Reinforce that Instagram/Meta does not handle credential verification through direct email correspondence.
    • Monitor for suspicious outbound communications to unfamiliar domains.
    • Encourage fast reporting of unusual security alerts to security teams.

    IOCs & Infrastructure:

    • Suspicious sender domains observed: @vacasa[.]uk.com, @syntec[.]uk.com, @pdftools[.]com.de, @boss[.]eu.com, among others.
    • Multiple malicious domains hosted together on shared IP infrastructure.

    Detection & Monitoring Recommendations:

    • Expand filters to flag alerts urging users to respond via email rather than through in-app or known support channels.
    • Track domain reputation for sender addresses linked to brand impersonation.
    • Focus awareness campaigns on the risks of replying to suspicious security notifications.

    This campaign highlights how attackers adapt by shifting tactics into spaces that feel safe and familiar. Defenses must adjust to catch threats that blend into ordinary communication.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    6 min
  • How Hackers Used Social Media to Deploy Cobalt Strike

    "Send me a quick text"

    Attackers abused phishing emails carrying shortcut files inside archives to deploy a legitimate crash-reporting executable paired with a malicious library. The library hijacked normal functions, retrieved encoded payloads hidden in GitHub, Quora, and Microsoft Tech Community profiles, and then redirected the victim system to GitHub raw content pages hosting encrypted shellcode. Once decrypted, the shellcode injected Cobalt Strike Beacon into memory, giving attackers full command-and-control inside the network.

    Defensive Actions

    • Monitor executables loading unsigned or unusual libraries.
    • Inspect outbound traffic to trusted platforms for encoded or repetitive fetches.
    • Detect reflective injection and executables that relaunch with hidden parameters.
    • Harden phishing defenses with archive and shortcut file scanning.

    Key IOCs

    • C2: moeodincovo[.]com/divide/mail/SUVVJRQO8QRC
    • Hosting platforms abused: GitHub repositories, Quora profiles, Microsoft Tech Community pages, Russian social networks

    Detection & Monitoring Focus

    • Hunt for abnormal library loads in crash-reporting or diagnostic executables.
    • Flag repeated HTTP requests to social media or developer sites fetching encoded data.
    • Track reflective memory injection techniques often tied to Cobalt Strike.

    Tools & Infrastructure

    • Legitimate crash-reporting utility hijacked with a malicious library
    • Trusted platforms (GitHub, Quora, Microsoft Tech Community) abused as staging points
    • Final payload: Cobalt Strike Beacon

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    7 min
  • Hijacking Microsoft Copilot AI

    "Send me a quick text"

    Episode Description – Technical Write-up for Defenders

    EchoLeak is the first documented zero-click AI vulnerability in a major enterprise LLM application — Microsoft 365 Copilot.
    The attacker seeds malicious instructions inside a normal-looking email, bypassing Microsoft’s prompt-injection filters.

    When a user later asks a related question, Copilot retrieves that email, executes the hidden commands, and packages sensitive corporate data for exfiltration. Bypassing link redaction and CSP rules, the attacker silently sends the data out through trusted Microsoft services.
    Persistence is achieved through “RAG spraying,” embedding the malicious instructions across multiple semantic chunks to maximize retrieval chances over time.

    Defensive Actions
    • Test AI prompt-injection classifiers, link filtering, and CSP enforcement in combination, not isolation.
    • Monitor AI assistant retrieval logs for unusual cross-context content use.
    • Implement strict scope enforcement so untrusted content cannot trigger privileged actions.
    • Scan indexed content stores for embedded malicious instructions or unusual markdown patterns.
    • Block unauthorized retrievals from trusted internal services to external destinations.

    Potential IOCs

    • Emails containing hidden markdown reference-style links
    • External requests routed through Microsoft media or rendering services
    • AI retrievals from HR, onboarding, or FAQ content containing unrelated data instructions

    Recommended Detection Focus

    • Alerts on Copilot responses containing embedded URLs or encoded data
    • Monitoring outbound requests from Microsoft services carrying sensitive information
    • Content scanning for prompt injection patterns inside indexed data

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    8 min
  • When Discord Becomes the Backdoor

    "Send me a quick text"

    Episode Description – Defender Briefing

    TMPN Stealer, derived from the open-source “Skuld” project, is an information-stealing malware that abuses Discord webhooks and client-side script injection to exfiltrate credentials, tokens, and 2FA codes. Once installed, it persists in the system, hides its activity, and bypasses security controls to maintain access. The attacker leverages Discord’s trusted infrastructure to blend exfiltration traffic with normal user activity, making detection difficult.

    Defensive Focus
    Monitor for unauthorized modifications to Discord client files, unexpected outbound traffic to Discord webhook endpoints, and abnormal Discord API activity (token validation, 2FA code usage). Review startup registry entries for suspicious “Realtek HD Audio Universal Service” values pointing to %APPDATA%\Microsoft\Protect\SecurityHealthSystray.exe. Apply strict endpoint controls to prevent unauthorized script execution within application directories.

    Known IOCs
    File: loader.exe
    SHA256: 5a7e38a45533e0477c3868c49df16d307a3da80b97a27ac4261619ff31a219f8
    Registry: HKCU\Software\Microsoft\Windows\CurrentVersion\Run\Realtek HD Audio Universal Service
    Network:

    • https://discord.com/api/webhooks/1272963856322527274/...
    • https://raw.githubusercontent.com/hackirby/discord-injection/main/injection.js
    • https://i.redd.it/68p07sk4976z.jpg

    Detection Tips
    Flag any system writing .js files into Discord’s appdata directories. Monitor PowerShell usage of Set-MpPreference with security feature disable flags. Inspect outbound Discord traffic for large JSON payloads or repeated authentication token validation requests.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    7 min
  • File Explorer's Hidden Danger

    "Send me a quick text"

    Episode Description – Technical Write-up for Defenders

    FileFix is a social engineering technique that builds on a known phishing method used in past ransomware and credential theft campaigns. It replaces more technical-looking prompts with a trusted and familiar Windows interface, making the request feel routine to the victim. By leveraging normal user actions inside a trusted application, the attacker can initiate malicious activity without relying on exploits or suspicious file downloads.

    Key Defensive Actions:

    • Restrict or limit script execution policies, especially for PowerShell.
    • Monitor for unusual process activity originating from trusted applications.
    • Provide awareness training that warns against unusual instructions involving system navigation tools.
    • Apply browser and email security controls to block interactions with untrusted domains.

    Relevant Indicators:

    • Unexpected process chains involving trusted applications and scripting engines.
    • Unusual clipboard activity linked to browser interactions.
    • Alerts for suspicious commands that originate outside of normal administrative workflows.

    Detection and Monitoring Tips:

    • Configure EDR/SIEM to flag scripting processes spawned by user-facing applications.
    • Enable detailed command-line auditing and review logs regularly.
    • Restrict high-risk browser functions when accessed from corporate devices.

    Known Infrastructure & Techniques:

    • Inspired by research into variations of the ClickFix phishing method.
    • Previously seen in campaigns attributed to ransomware operators and state-sponsored groups.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    7 min
  • Hacking AI with No Clicks

    "Send me a quick text"

    Episode Description:

    AgentFlayer is a series of AI-targeted exploits that use language itself as the payload. By embedding malicious instructions inside everyday documents or text, such as a colleague’s CV, an industry report shared on LinkedIn, or any file that “needs summarizing,” attackers can trigger AI assistants to exfiltrate sensitive data. This includes stored memory, account metadata, and even entire conversation histories, sometimes without the victim clicking anything at all. The method blends perfectly with normal AI workflows, bypassing traditional security controls.

    Defensive Actions:

    • Treat all external documents and pasted text as untrusted before sending them to AI systems.
    • Implement AI input validation and sanitization to strip or neutralize hidden prompt instructions.
    • Monitor AI-assisted workflows for unusual outbound requests or data patterns.
    • Restrict AI system access to sensitive data and stored memory where possible.
    • Educate users on the risks of summarizing unverified content, even from trusted networks.

    IOCs & Monitoring Focus:

    • Look for unexpected outbound connections from AI-related services or integrations.
    • Monitor logs for unusual API calls or data retrieval requests triggered immediately after file uploads.
    • Track access to AI “memory” or conversation archives outside normal user behavior.
    • Investigate any requests from AI systems that direct users to log in on unfamiliar domains.

    Known Tools / Infrastructure:

    • Exploitation is performed entirely via indirect prompt injection.
    • May involve attacker-controlled domains hosting phishing pages.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    6 min
  • The Malware That Fights Back

    "Send me a quick text"

    Episode Description

    XWorm 6.0 adds a hostile twist to persistence. Delivered via obfuscated VBScript, it bypasses AMSI, runs entirely in memory, and marks itself as a “critical process” that crashes your PC if you try to kill it. In this episode, we walk through the attack from delivery to defense — and why it works so well.

    Technical Write-Up

    Initial Access: Delivered via obfuscated VBScript, likely through social engineering. Payload concealed as reversed numeric character codes, reconstructed at runtime.

    Execution: Removes Zone.Identifier ADS, launches PowerShell to download a secondary script into the TEMP folder.

    Persistence: Copies itself as update.vbs into TEMP and APPDATA, adds both to Windows Run registry key. Builder also supports scheduled tasks or startup folder persistence.

    AMSI Bypass: Patches clr.dll in memory, replacing “AmsiScanBuffer” with null bytes to disable AMSI memory inspection.

    Payload Delivery: Downloads XWorm binary from public GitHub repository via .NET HTTP client. Loads into memory with Assembly.Load for fileless execution.

    Privilege Escalation & Process Protection: With admin rights, enables SeDebugPrivilege and marks itself as a “critical process,” causing a system crash if terminated.

    Anti-Analysis: Terminates on Windows XP and if IP-API lookup detects hosting or data center IP ranges.

    Defensive Recommendations:

    • Monitor for unusual VBScript execution, especially from email or downloads.
    • Detect PowerShell downloading scripts from public repositories.
    • Alert on in-memory modifications to clr.dll.
    • Investigate registry Run keys pointing to TEMP or APPDATA.
    • Watch for processes setting themselves as “critical.”
    • Track outbound requests to IP-API or similar services.

    Indicators of Compromise (IOCs) & Scripts:
    All the IOCs and scripts related to this malware are available in the Netskope Threat Labs GitHub repository:
    https://github.com/netskopeoss/NetskopeThreatLabsIOCs/tree/main/Malware/XWorm/IOCs/2025-07-25

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    7 min
  • Cracking the Fingerprint Vault

    "Send me a quick text"

    The Dell ControlVault3 exploit targets the firmware of the biometric security chip found in many Dell laptops. By injecting malicious code into the firmware — either through local OS-level access or direct physical connection — attackers gain persistent, hardware-level control. This implant can survive OS reinstalls, grant SYSTEM privileges, and remain dormant until triggered by specific biometric commands.

    Why It’s Dangerous

    • Stealth Layer — Runs below the operating system, invisible to most endpoint security tools.
    • Persistence — Survives disk wipes and OS reinstalls; removal requires reflashing or replacing the hardware.
    • Hardware Trust Abuse — Exploits the fact that the OS fully trusts ControlVault3’s responses.

    Detection Strategy

    • Firmware Integrity Monitoring — Use tools like Intel CHIPSEC to compare the current firmware against vendor-signed versions.
    • Event Logging — Enable auditing for firmware update events, especially via internal USB or service ports.
    • Behavioral Anomalies — Alert on rare or suspicious biometric commands, particularly those outside normal authentication flows.

    Prevention Strategy

    • Firmware Attestation — Implement secure boot and cryptographic verification for firmware images before applying updates.
    • Access Control — Restrict firmware flashing utilities to authorized admin accounts and protected service environments.
    • Physical Security — Enforce tamper-evident seals, controlled laptop storage, and secure transport procedures for high-risk devices.

    Relevant Files / Interfaces

    • ControlVault3 firmware images (*.bin, *.fw).
    • Internal USB interface for service access.
    • Firmware update utilities (dvupdate.exe or similar).

    Recommended Tools

    • CHIPSEC — For low-level firmware integrity validation.
    • Dell Command Update with Policy Control — To enforce vendor-signed firmware updates only.
    • SIEM/EDR Integration — Create detection rules for unusual firmware update activity and biometric command anomalies.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    8 min
  • The Invisible Intruder: Deconstructing LockBit

    "Send me a quick text"

    This episode explores a LockBit ransomware campaign that relied on DLL sideloading and masquerading to operate undetected until the final encryption stage.

    The attackers gained access using legitimate remote management tools already present in the environment. They paired trusted, signed applications with malicious DLLs placed in locations where the application would load them first. Masquerading techniques — such as adopting common process names, using standard icons, and placing payloads in system-like directories — allowed their presence to blend seamlessly into normal operations.

    Once inside, they escalated privileges, conducted network reconnaissance, stole credentials and Kerberos tickets, and used Group Policy to distribute payloads. The ransomware was ultimately launched under the identity of a trusted process, bypassing many traditional detection points.

    Defensive Recommendations

    • Restrict DLL search paths and enable Safe DLL Search Mode to prevent loading from user-writable or non-standard directories.
    • Deploy application control policies (e.g., Windows Defender Application Control, AppLocker) to block unapproved binaries and DLLs.
    • Monitor for legitimate applications loading DLLs from unusual directories or with unexpected hashes.
    • Audit the use of remote management tools; require MFA, logging, and alerting for unexpected file transfers or executions.
    • Track Group Policy changes, focusing on script and binary deployments to endpoint systems.
    • Monitor for obfuscated PowerShell commands that interact with large sets of files or sensitive directories.

    Files, Folders, Tools, and Configurations for Defenders

    • Files and Folders: Watch for legitimate application executables paired with DLLs of the same name in non-standard or writable directories; payloads appearing in ProgramData or user profile subdirectories.
    • Tools: Remote access tools such as TeamViewer and MeshAgent; service management utilities like NSSM; administrative execution tools like PsExec; and credential theft tools for token manipulation and Kerberos extraction.
    • Configurations:
      • Enforce strict application allowlisting for both executables and DLLs.
      • Limit write permissions to directories used by trusted applications.
      • Enable command-line logging for PowerShell, PsExec, and similar tools.
      • Implement centralized logging and correlation rules to detect legitimate applications initiating encryption or credential theft activity.

    Thanks for spending a few minutes on the CyberBrief Project.

    If you want to dive deeper or catch up on past episodes, head over to cyberbriefproject.buzzsprout.com.

    You can also find the podcast on YouTube at youtube.com/@CyberBriefProject — I’d love to see you there.

    And if you find these episodes valuable and want to support the project, you can do that here: buzzsprout.com/support

    Your support means a lot.

    See you in the next one, and thank you for listening.

    11 min

About CyberBrief Project

From the publisher's feed

CyberBrief Project is an audio series that breaks down one creative cyber-attack technique in each episode.


Rather than covering routine threats, each episode focuses on clever…