Welcome to this week’s version of the GBHackers cybersecurity publication — your weekly cybersecurity bulletin protecting the 40 most necessary tales from July 6–10, 2026.
This week the safety world collided with AI head-on: prompt-injection assaults turned chatbots into C2 brokers, 5 main AI coding assistants fell to a single assault sample, and hundreds of MCP servers have been discovered uncovered.
In the meantime, CitrixBleed 2 exploitation went from idea to DragonForce ransomware deployments, Android 17 bought rooted with one click on, and a ransomware negotiator went to jail for enjoying each side. Right here’s every part your friends are studying this week.
IN THIS ISSUE
High Tales of the Week — 5 tales
AI Below Assault — 7 tales
Important Vulnerabilities & Exploits — 11 tales|
Ransomware & Extortion — 5 tales
Phishing & Social Engineering — 6 tales
Breaches, Malware & Provide Chain — 6 tales
🔥 TOP STORIES OF THE WEEK
1. Hackers Exploit CitrixBleed 2 to Hijack MFA-Protected Periods and Deploy DragonForce Ransomware
July 10, 2026 • gbhackers.com
Menace actors are exploiting CVE-2025-5777 to hijack energetic NetScaler classes — even these protected by multi-factor authentication. As soon as inside, they’re transferring laterally and deploying DragonForce ransomware throughout enterprise networks.
July 9, 2026 • gbhackers.com
Researchers present Claude Desktop’s synced Private Preferences function might be abused as a covert prompt-injection channel, turning the assistant right into a de facto command-and-control agent. The approach chains all the way in which to distant code execution on the sufferer’s machine.
3. 281 Android VPN Apps Expose Customers to Visitors Leaks, Monitoring and Tunnel Hijacking
July 10, 2026 • gbhackers.com
A sweeping safety evaluation discovered 281 Android VPN apps exposing customers to visitors leaks, third-party monitoring, and tunnel hijacking. The instruments thousands and thousands put in for privateness could be the largest privateness threat on the telephone.
4. IonStack Exploit Chain Lets Hackers Root Android 17 Telephones With a Single URL Click on
July 8, 2026 • gbhackers.com
Nebula Safety’s “IonStack” exploit chain achieves full root entry on Android 17 gadgets from a single malicious URL click on. Even Google’s latest, most hardened Android construct can fall to at least one faucet.
5. Ransomware Negotiator Jailed for Leaking Sufferer Secrets and techniques to BlackCat Hackers
July 10, 2026 • gbhackers.com
A Florida ransomware negotiator has been sentenced to federal jail for conspiring with ALPHV/BlackCat operators whereas supposedly negotiating on victims’ behalf. He was quietly making the most of each side of the extortion desk.
🤖 AI UNDER ATTACK
6. GhostApproval Assault Impacts Amazon Q, Claude Code, Cursor, Google Antigravity, and Windsurf
July 9, 2026 • gbhackers.com
A newly disclosed vulnerability sample dubbed “GhostApproval” breaks the human-approval belief boundary on the coronary heart of AI coding assistants. Amazon Q, Claude Code, Cursor, Google Antigravity, and Windsurf are all affected.
7. GitHub Copilot IDE Coding Brokers Susceptible to Workflow-Degree Jailbreak Assaults
July 9, 2026 • gbhackers.com
GitHub Copilot’s new IDE coding brokers might be jailbroken on the workflow degree, bypassing the guardrails constructed into the editor. Attackers can steer the agent into actions the developer by no means permitted.
8. Google Gemini Reside API Flaw Permits RCE through Unconstrained Ephemeral Tokens
July 7, 2026 • gbhackers.com
Misconfigured ephemeral tokens in Google’s Gemini Reside API left integrating functions open to distant code execution. Researchers present how unconstrained tokens hand attackers way more energy than builders meant.
9. Hundreds of MCP Servers Discovered Susceptible to File Entry and Injection Assaults
July 7, 2026 • gbhackers.com
Hundreds of Mannequin Context Protocol servers — the connectors linking LLMs to exterior programs — are weak to important file entry and injection assaults. The AI ecosystem’s fastest-growing plumbing can also be its least secured.
10. Hackers Compromise AWS AI Gateway Related to Amazon Bedrock to Deploy XMRig Cryptominer
July 10, 2026 • gbhackers.com
Attackers compromised an enterprise AI gateway linked to Amazon Bedrock and used it to deploy an XMRig cryptominer. Generative AI infrastructure has formally joined the enterprise assault floor.
11. US Cyber Company Makes use of Anthropic Mythos to Audit Authorities Code for Bugs
July 7, 2026 • gbhackers.com
CISA has begun utilizing Anthropic’s superior Mythos mannequin to audit authorities software program for vulnerabilities. It marks the primary large-scale federal deployment of a frontier AI mannequin for code safety evaluate.
July 10, 2026 • gbhackers.com
Microsoft is increasing a multi-model “agentic” AI scanning system that hunts Home windows safety flaws earlier than attackers discover them. The corporate says it’ll meaningfully speed up patch supply worldwide.
⚠️ CRITICAL VULNERABILITIES & EXPLOITS
13. Microsoft Edge Excessive-Severity Vulnerability Permits Distant Code Execution
July 7, 2026 • gbhackers.com
Microsoft has disclosed CVE-2026-57992, a high-severity distant code execution flaw in its Chromium-based Edge browser. A crafted web page might hand attackers management of unpatched programs — replace instantly.
14. Linux FUSE Vulnerability Permits Unprivileged Customers to Pop a Root Shell
July 10, 2026 • gbhackers.com
A newly disclosed flaw within the Linux kernel’s FUSE subsystem lets unprivileged native customers corrupt the web page cache and escalate straight to root. Any shared or multi-user Linux system is within the blast radius.
15. Dangerous Epoll Linux Kernel UAF Flaw Lets Unprivileged Attackers Achieve Root on Linux and Android
July 6, 2026 • gbhackers.com
CVE-2026-46242, dubbed “Dangerous Epoll,” is a race-condition use-after-free within the kernel’s epoll subsystem. It fingers unprivileged attackers root on each Linux servers and Android gadgets.
16. 16-Yr-Outdated Januscape KVM Escape Vulnerability Lets Attackers Compromise Linux Hosts
July 7, 2026 • gbhackers.com
CVE-2026-53359, named “Januscape,” exposes a flaw that sat in KVM/x86 virtualization for 16 years. A malicious visitor VM can escape and absolutely compromise the underlying Linux host.
17. Veeam Backup BinaryFormatter Flaw Allows Distant Code Execution
July 6, 2026 • gbhackers.com
CVE-2026-44963 is a BinaryFormatter deserialization flaw in Veeam Backup & Replication that lets authenticated area customers execute distant code. Backup servers are precisely the goal ransomware crews hit first.
18. Important BeyondTrust Authentication Flaws Expose Distant Assist Home equipment to Assaults
July 7, 2026 • gbhackers.com
BeyondTrust disclosed a number of important and high-severity flaws in its Distant Assist and Privileged Distant Entry home equipment. The instruments constructed to protect privileged entry have change into the doorway in.
19. Roundcube Webmail Safety Replace Patches Important Zero-Click on XSS and SSRF Bypass Flaws
July 10, 2026 • gbhackers.com
Roundcube 1.7.2 patches a zero-click saved XSS that fires with no consumer interplay in any respect, plus an SSRF bypass. Self-hosted webmail admins ought to deal with this as an emergency replace.
20. Important Opera GX Vulnerability Lets Attackers Inject CSS Throughout Each Webpage
July 6, 2026 • gbhackers.com
A important flaw in Opera GX’s Mods function allowed attackers to inject malicious CSS into each webpage a consumer visits. A function constructed for browser customization grew to become a common injection channel.
21. Over 70% of Public WordPress Websites Operating Outdated PHP Uncovered to Cyberattacks
July 8, 2026 • gbhackers.com
New evaluation reveals greater than 70% of publicly accessible WordPress websites run outdated, unsupported PHP variations. That’s tens of thousands and thousands of websites uncovered to identified, already-weaponized exploits.
22. Attackers Exploit WordPress Plugin Vulnerabilities for Distant Code Execution and Webshell Entry
July 9, 2026 • gbhackers.com
A big-scale marketing campaign is actively weaponizing identified WordPress plugin flaws for distant code execution and webshell deployment. Unpatched plugins stay the one easiest method into the CMS ecosystem.
23. Wireshark 4.6.7 Launched to Patch 12 Vulnerabilities in SSH, TLS, Wi-Fi and pcapng
July 10, 2026 • gbhackers.com
Wireshark 4.6.7 fixes 12 safety flaws throughout SSH, TLS and Wi-Fi dissectors and pcapng file parsing. A booby-trapped seize file might compromise the very analyst investigating an incident.
🔒 RANSOMWARE & EXTORTION
July 9, 2026 • gbhackers.com
Technical evaluation of an Everest encryptor reveals a closely obfuscated, ConfuserEx-protected .NET binary with built-in Wake-on-LAN functionality. The malware actually wakes sleeping machines on the community so it could encrypt them too.
July 9, 2026 • gbhackers.com
The “GodDamn” ransomware seems to be the newest rebrand of a long-running household reasonably than one thing new. Its operators lean on PsExec for lateral motion and NirSoft utilities for credential theft.
26. New Helix Extortion Group Targets Enterprises With MFA Abuse and SharePoint Exfiltration
July 9, 2026 • gbhackers.com
A beforehand unreported extortion crew dubbed “Helix” is hitting enterprises with voice phishing, MFA abuse, and automatic SharePoint knowledge theft. It’s identity-first extortion — no encryptor required.
27. TeamPCP Provide Chain Assaults Feed VECT Ransomware With Stolen CI/CD Credentials
July 7, 2026 • gbhackers.com
TeamPCP’s wide-scale supply-chain compromises have amassed an unlimited archive of stolen CI/CD credentials now straight fueling VECT ransomware operations. Your pipeline secrets and techniques could already be of their fingers.
28. GigaWiper Makes use of OneDrive Replace Scheduled Activity for Persistent Harmful Entry
July 10, 2026 • gbhackers.com
GigaWiper is a Golang-based backdoor that hides its persistence inside a scheduled job disguised as a OneDrive replace. Behind its intensive C2 controls sit a number of damaging wiper payloads awaiting the set off.
🎣 PHISHING & SOCIAL ENGINEERING
29. Forg365 PhaaS Makes use of Telegram and AI Lures to Hijack Microsoft 365 Accounts
July 10, 2026 • gbhackers.com
Forg365 is a business phishing-as-a-service platform aimed squarely at Microsoft 365, combining device-code phishing, adversary-in-the-middle workflows, and AI-assisted lures. The entire operation is run and bought by means of Telegram.
30. Faux Robinhood Signal-In Alerts Trick Customers Into Calling Hacker-Managed Telephone Numbers
July 10, 2026 • gbhackers.com
A callback-phishing wave is sending faux Robinhood “new sign-in” alerts that stress customers into dialing attacker-controlled telephone numbers. The panic is manufactured; the account takeover that follows is actual.
31. Hackers Use Trusted Microsoft Area and One-Time Codes to Hijack Company Accounts
July 6, 2026 • gbhackers.com
Attackers are weaponizing a official Microsoft OAuth 2.0 stream and one-time codes to take over company accounts — no stolen passwords required. As a result of the lure lives on a trusted Microsoft area, electronic mail filters wave it straight by means of.
32. Hackers Abuse Cross-Tenant Groups Chat to Ship EtherRAT By Malicious MSI Loader
July 7, 2026 • gbhackers.com
A marketing campaign noticed in late June pairs electronic mail phishing with abused cross-tenant Microsoft Groups chats to ship the delicate EtherRAT through a malicious MSI loader. Exterior Groups messages are the brand new phishing inbox.
33. SNOW Malware Ecosystem Makes use of Groups Phishing, WebSocket Tunnels, and Browser Extensions
July 9, 2026 • gbhackers.com
The SNOW malware ecosystem chains convincing Groups phishing, covert WebSocket tunnels, and rogue browser extensions into one plausible intrusion path. It’s a full-stack social-engineering operation, not a single payload.
34. New Multi-Stage LNK Assault Targets Hospitality Companies With Node.js Backdoor
July 10, 2026 • gbhackers.com
Accommodations and journey companies are being hit with faux booking-related emails that set off a multi-stage LNK assault chain. The ultimate payload is a Node.js backdoor giving attackers full distant management.
🦠 BREACHES, MALWARE & SUPPLY CHAIN
July 9, 2026 • gbhackers.com
US auto and renters insurer AssuranceAmerica confirmed a major breach exposing clients’ private data and driver’s license knowledge. Affected policyholders now face elevated identity-theft and fraud threat.
36. npm and PyPI Malware Marketing campaign Exfiltrates CI/CD Secrets and techniques By Faux Cost SDKs
July 9, 2026 • gbhackers.com
A coordinated marketing campaign pushed 17 malicious npm and PyPI packages masquerading as SDKs for PaySafe, Skrill, and different fee providers. Their actual job: silently exfiltrating CI/CD secrets and techniques from construct environments.
37. RedHook Abuses Accessibility Service to Allow Developer Choices and Wi-fi Debugging
July 9, 2026 • gbhackers.com
The RedHook Android RAT has resurfaced with the flexibility to autonomously allow Developer Choices and wi-fi debugging by abusing accessibility providers. That unlocks a far deeper degree of machine management than earlier than.
38. Lurking Lizard Makes use of Drop-Catch Domains and Lookalike Manufacturers to Distribute Proxyware
July 8, 2026 • gbhackers.com
The long-running “Lurking Lizard” operation makes use of drop-catch domains and lookalike manufacturers — together with a faux 7-Zip installer — to unfold proxyware. Contaminated PCs are silently rented out as residential proxy exit nodes.
39. ESET Menace Report H1 2026 Highlights Malicious AI Expertise, ClickFix Surge, and EDR Killers
July 8, 2026 • gbhackers.com
ESET’s first-half 2026 report flags malicious AI abilities, a surging ClickFix approach, and more and more widespread EDR-killer tooling. Attackers aren’t inventing new playbooks — they’re supercharging outdated ones with AI.
40. Malicious Braintree.Web Typosquat Steals PAN, CVV, and Cost Gateway Credentials
July 10, 2026 • gbhackers.com
A malicious NuGet bundle typosquatting Braintree’s official .NET shopper was caught stealing card numbers, CVVs, and fee gateway credentials. Automated scanning flagged it as malware inside minutes of its July 3 publication.
❓ FREQUENTLY ASKED QUESTIONS
What does this weekly cybersecurity publication cowl?
Every problem of the GBHackers cybersecurity publication rounds up the week’s 40 most necessary tales — important vulnerabilities, ransomware assaults, knowledge breaches, AI safety threats, phishing campaigns, and malware analysis — curated by our editorial staff from every part printed on gbhackers.com.
How is a cybersecurity bulletin totally different from every day safety information?
A cybersecurity bulletin condenses lots of of every day headlines right into a single prioritized weekly briefing. As a substitute of monitoring feeds all day, safety groups get the exploited CVEs, energetic campaigns, and breaches that really matter — with direct hyperlinks to the complete evaluation of every story.
How do I subscribe to the GBHackers weekly cybersecurity publication?
Go to gbhackers.com and observe us on LinkedIn or X (@gbhackers_news) to get each weekly problem. The publication is free and lands as soon as every week, each week.
Discovered this cybersecurity bulletin helpful? Get the weekly cybersecurity publication in your inbox — free, each week, from GBHackers.
