Major Incidents or Breaches

  • Hackers exploited a SQL injection vulnerability in a public-facing web application to gain access to an Oracle database, then compiled and ran the khunt post-exploitation toolkit inside the database, enabling SYSTEM-level access without writing executables to disk [1][19].
  • Brown Health Medical Group-MA suffered a data breach impacting 311,000 individuals, with hackers stealing personal, medical, and financial information from the organization’s server [48].
  • A phishing campaign exploited concerns over a COLDCARD wallet vulnerability and alleged Bitcoin theft to trick users into installing the ScreenConnect remote access tool [20].
  • GitGuardian researchers identified 321 n8n automation instances exposed due to leaked API tokens in public GitHub commits, allowing credential theft and access to sensitive data [16].
  • A Canadian national pleaded guilty to hacking Snowflake cloud storage customer accounts, stealing data from at least 165 organizations and affecting at least 100 million people [6][18].

Newly Discovered Vulnerabilities

  • A critical remote code execution vulnerability (CVE-2026-63077) in JetBrains TeamCity is under active exploitation, allowing unauthenticated attackers to execute code on on-premise servers [5][43].
  • A memory corruption flaw in the Linux kernel’s Open vSwitch datapath allows local users to escalate privileges to root across multiple default-configured distributions, with a public exploit available [13].
  • Cisco released patches for critical vulnerabilities in SD-WAN, IOS XE, and FMC products, including one with publicly available proof-of-concept code [42].
  • HashiCorp Terraform MCP Server, Veeam Service Provider Console, and Django patched 11 vulnerabilities, including a CVSS 10.0 cross-tenant bug in Terraform MCP that allows unauthenticated access to sensitive data [11].
  • A critical flaw in Gitea (versions 1.22.1 through 1.27.0) allows unauthenticated attackers to read any server file accessible to the service account via Org-Mode markup, with a public read exploit released [15].
  • At least 20 models of Zbtlink routers were found to ship with a factory-installed backdoor, permitting unauthenticated root shell access [3].
  • Security flaws in agent infrastructure from AWS, Google, and Vercel allowed untrusted or forged instructions to reach agent tools without proper authorization checks [2].
  • Two Paperclip AI control plane vulnerabilities allow attackers to execute arbitrary commands on network servers or developer machines via malicious agent imports [10].
  • Flaws in the Google APK for Python permitted agent-to-agent attacks by exploiting trust boundaries between AI agents with differing privilege levels [29].
  • Google’s synchronized passkeys are vulnerable to malware-based “Pass-ta-key” attacks, allowing theft of master keys and hijacking of passkey-protected accounts [47][50].
  • Researchers disclosed 15 vulnerabilities in TP-Link devices, highlighting risks in automated zero-trust provisioning [28].

Notable Threat Actor Activity

  • Kali365, a phishing kit operator, is targeting US organizations by weaponizing Microsoft authentication flows, using attacker-controlled device codes that victims approve to gain access to corporate data [14].
  • Over 250 domains associated with the ClickFix macOS malware campaign now use browser fingerprinting to selectively display malware lures, evading detection [7].
  • A trojanized npm package campaign employs the NullReceiver tactic, decoding command-and-control server IP addresses from blockchain transactions to evade traditional detection and takedown efforts [12].
  • Multiple underground services are selling discounted access to AI models such as Claude, with operators able to view all customer prompts, raising concerns about data exposure [9].
  • OpenAI disrupted a Cambodia-based scam network using ChatGPT to facilitate a range of fraud schemes, including investment, romance, gambling, and law enforcement scams [8].
  • A security researcher maintained access to North Korean hacking infrastructure for nearly two years, revealing that North Korean actors breached hundreds of networks worldwide [32].

Trends, Tools, or Tactics of Interest

  • AI-powered phishing campaigns are rendering traditional blocklists ineffective, as attackers rapidly create disposable infrastructure and evolve toolkits faster than blocklists can respond; browser-level and technique-based detection are recommended [23].
  • Organized crime groups are leveraging AI technologies such as voice cloning, deepfake video overlays, LLM-driven persona management, and automated translation for large-scale fraud operations [24].
  • Prompt injection and agent hijacking vulnerabilities remain prevalent across leading AI browsers and agent platforms, with no simple fixes; attackers can hijack agents via malicious content or instructions [25][26][31].
  • CSS can be abused to exfiltrate data from webmail, representing a growing threat vector as some vendors remain unprepared for such attacks [27].
  • Automated SSH brute-force actors can compromise and persist on exposed systems in as little as 22 seconds, highlighting the speed and automation of current attack techniques [38].
  • New attack methods demonstrated that malware can hijack Google’s passkey-protected accounts via synced passkey implementation, bypassing passwordless security [47][50].
  • The EtherHiding blockchain-based C2 technique continues to evolve, with attackers using fabricated blockchain addresses to conceal C2 IPs in npm package campaigns [12].

Regulatory or Policy Developments Affecting the Security Industry

  • The Open Secure AI Alliance, now comprising 120 organizations, has drafted SAFE guidelines to standardize sharing of AI incident data [49].
  • The US Cybersecurity and Infrastructure Security Agency (CISA) issued a directive for federal agencies to mitigate actively exploited vulnerabilities in JetBrains TeamCity, IBM Langflow, N-central, and Apache Tomcat within three days [5][21][43].
  • Maksim Silnikau, creator and administrator of the Ransom Cartel ransomware-as-a-service operation, was sentenced to 16 years in prison for attacks against at least 18 organizations [4][17][41].

[1]: Attackers Compile khunt Inside Oracle to Turn SQL Injection Into Windows SYSTEM Access
[3]: Chinese-Made Zbtlink Routers Ship With Backdoor That Opens Unauthenticated Root Shells
[4]: Ransom Cartel Creator Gets 16 Years in Prison for Operating Ransomware-as-a-Service
[5]: CISA Flags TeamCity CVE-2026-63077 RCE Flaw Under Active Exploitation in the Wild
[6]: Snowflake Hacker Pleads Guilty Over Breaches Affecting at Least 100 Million People
[7]: Over 250 ClickFix Domains Use Browser Fingerprinting to Hide macOS Malware Lures
[8]: OpenAI Disrupts Poipet Scam Network Using ChatGPT Across Multiple Fraud Schemes
[9]: Poison Claude Sells Discounted Claude Access While Its Operator Sees Every Customer Prompt
[10]: Paperclip AI Flaws Let Attackers Run Host Commands via Malicious Agent Imports
[11]: Veeam, Terraform MCP, Django Patch Critical Flaws, Led by CVSS 10.0 Cross-Tenant Bug
[12]: Trojanized npm Packages Employ NullReceiver Tactic to Decode C2 IP from Blockchain
[13]: New OVSwrap Linux Kernel Flaw Lets Local Users Gain Root via Open vSwitch
[14]: Kali365 Weaponizes Microsoft Authentication Against US Companies: New Enterprise Risk
[15]: Critical Gitea Flaw Let Unauthenticated Attackers Read Server Files via Org-Mode Markup
[16]: Leaked n8n API Tokens Exposed Live Instances to Credential Theft
[17]: Ransom Cartel ransomware creator sentenced to 16 years in prison
[18]: Canadian pleads guilty to Snowflake cloud data-theft attacks
[19]: Hackers run khunt post-exploitation toolkit from Oracle database
[20]: COLDCARD security audit phishing attack installs remote access tool
[21]: CISA warns of hackers exploiting Langflow, N-central, Apache Tomcat flaws
[23]: How AI-powered phishing killed blocklists for good
[24]: AI Sends Global Crime Syndicates Into Fraud Nirvana
[25]: AI Browsers Vulnerable to ‘PleaseFix’ Zero-Click Agent Hijacking
[26]: No Perfect Fix for AI Browser Prompt Injection Flaws
[27]: CSS: The Hidden Threat Lurking in Your Inbox
[28]: 15 TP-Link Bugs Expose Risks in Zero-Trust Provisioning
[29]: Flaws in Google APK for Python Unlock Agent-to-Agent Attack
[31]: OpenAI’s Browser Could Be Hijacked to Spam Your WhatsApp Contacts
[32]: A Security Pro Hacked North Korean Hackers. He Found They’d Breached Hundreds of Networks Worldwide
[38]: 22 Seconds to Compromise: How Automated SSH Actors Move From Login to Persistence Before You Can Blink
[41]: Belarusian Ransom Cartel Mastermind Gets 16 Years in Prison
[42]: Cisco Patches Critical SD-WAN, IOS XE, FMC Vulnerabilities
[43]: Hackers Start Exploiting Recent JetBrains TeamCity Vulnerability
[47]: New Attack Methods Enable Malware to Hijack Passkey-Protected Accounts
[48]: 311,000 Impacted by Brown Health Medical Group-MA Data Breach
[49]: Cybersecurity Alliance Drafts SAFE Guidelines for Sharing AI Incident Data
[50]: Google’s synchronized passkeys can be stolen in ‘Pass‑ta‑key’ attacks

Sources#

  1. Attackers Compile khunt Inside Oracle to Turn SQL Injection Into Windows SYSTEM Access - thehackernews.com
  2. Hackers run khunt post-exploitation toolkit from Oracle database - bleepingcomputer.com
  3. 311,000 Impacted by Brown Health Medical Group-MA Data Breach - securityweek.com
  4. COLDCARD security audit phishing attack installs remote access tool - bleepingcomputer.com
  5. Leaked n8n API Tokens Exposed Live Instances to Credential Theft - thehackernews.com
  6. Snowflake Hacker Pleads Guilty Over Breaches Affecting at Least 100 Million People - thehackernews.com
  7. Canadian pleads guilty to Snowflake cloud data-theft attacks - bleepingcomputer.com
  8. CISA Flags TeamCity CVE-2026-63077 RCE Flaw Under Active Exploitation in the Wild - thehackernews.com
  9. Hackers Start Exploiting Recent JetBrains TeamCity Vulnerability - securityweek.com
  10. New OVSwrap Linux Kernel Flaw Lets Local Users Gain Root via Open vSwitch - thehackernews.com
  11. Cisco Patches Critical SD-WAN, IOS XE, FMC Vulnerabilities - securityweek.com
  12. Veeam, Terraform MCP, Django Patch Critical Flaws, Led by CVSS 10.0 Cross-Tenant Bug - thehackernews.com
  13. Critical Gitea Flaw Let Unauthenticated Attackers Read Server Files via Org-Mode Markup - thehackernews.com
  14. Chinese-Made Zbtlink Routers Ship With Backdoor That Opens Unauthenticated Root Shells - thehackernews.com
  15. AWS, Google, and Vercel Agent Flaws Let Attackers Trigger Tools Without Running the Model - thehackernews.com
  16. Paperclip AI Flaws Let Attackers Run Host Commands via Malicious Agent Imports - thehackernews.com
  17. Flaws in Google APK for Python Unlock Agent-to-Agent Attack - darkreading.com
  18. New Attack Methods Enable Malware to Hijack Passkey-Protected Accounts - securityweek.com
  19. Google’s synchronized passkeys can be stolen in ‘Pass‑ta‑key’ attacks - malwarebytes.com
  20. 15 TP-Link Bugs Expose Risks in Zero-Trust Provisioning - darkreading.com
  21. Kali365 Weaponizes Microsoft Authentication Against US Companies: New Enterprise Risk - thehackernews.com
  22. Over 250 ClickFix Domains Use Browser Fingerprinting to Hide macOS Malware Lures - thehackernews.com
  23. Trojanized npm Packages Employ NullReceiver Tactic to Decode C2 IP from Blockchain - thehackernews.com
  24. Poison Claude Sells Discounted Claude Access While Its Operator Sees Every Customer Prompt - thehackernews.com
  25. OpenAI Disrupts Poipet Scam Network Using ChatGPT Across Multiple Fraud Schemes - thehackernews.com
  26. A Security Pro Hacked North Korean Hackers. He Found They’d Breached Hundreds of Networks Worldwide - wired.com
  27. How AI-powered phishing killed blocklists for good - bleepingcomputer.com
  28. AI Sends Global Crime Syndicates Into Fraud Nirvana - darkreading.com
  29. AI Browsers Vulnerable to 'PleaseFix' Zero-Click Agent Hijacking - darkreading.com
  30. No Perfect Fix for AI Browser Prompt Injection Flaws - darkreading.com
  31. OpenAI’s Browser Could Be Hijacked to Spam Your WhatsApp Contacts - wired.com
  32. CSS: The Hidden Threat Lurking in Your Inbox - darkreading.com
  33. 22 Seconds to Compromise: How Automated SSH Actors Move From Login to Persistence Before You Can Blink [Guest Diary], (Thu, Aug 6th) - isc.sans.edu
  34. Cybersecurity Alliance Drafts SAFE Guidelines for Sharing AI Incident Data - securityweek.com
  35. CISA warns of hackers exploiting Langflow, N-central, Apache Tomcat flaws - bleepingcomputer.com
  36. Ransom Cartel Creator Gets 16 Years in Prison for Operating Ransomware-as-a-Service - thehackernews.com
  37. Ransom Cartel ransomware creator sentenced to 16 years in prison - bleepingcomputer.com
  38. Belarusian Ransom Cartel Mastermind Gets 16 Years in Prison - securityweek.com