🛑 Malicious AI skills can bypass scanners.

SKILLCLOAK hides payloads in folders scanners often skip, then rebuilds them only when the agent runs the skill.

In tests, its packing trick bypassed all 8 scanners more than 90% of the time.

Read how the cloaking works: https://thehackernews.com/2026/07/new-skillcloak-technique-lets-malicious.html

🚨 A zero-click Opera GX flaw let malicious sites silently install GX Mods and leak data exposed in pages the victim visited.

In a PoC, researchers used CSS to rebuild a signed-in user’s Gmail address after one visit.

Learn how “just styling” crossed sites: https://thehackernews.com/2026/07/opera-gx-flaw-let-malicious-sites-auto.html

⚠️ New Java RAT is being sold as MaaS.

QuimaRAT is built to run on Windows, #Linux, and #macOS, with encrypted plugins, OS-specific persistence, C2 rotation via Pastebin, and a browser-cache loader for staging payloads.

Read: https://thehackernews.com/2026/07/new-java-based-quimarat-maas-built-to.html

🚨 New TrojPix attack can leak data from an air-gapped PC through its video cable.

It uses user-level malware to turn invisible pixel changes into radio signals from a video cable, leaking data at up to 8.1 Mbps.

Learn how TrojPix turns pixels into a radio leak: https://thehackernews.com/2026/07/new-trojpix-attack-leaks-data-from-air.html

🚨 A suspected China-nexus campaign is targeting Indian taxpayers, tax professionals, and finance teams with fake Income Tax Department emails.

Operation DragonReturn leads victims to a fake tax-filing utility that uses DLL sideloading, JPG payload concealment, and DCRat.Attribution remains unclear.

Read: https://thehackernews.com/2026/07/suspected-china-nexus-hackers-use-fake.html

AI SOC can mean a chat box bolted onto a SIEM, or agents that handle detection, triage, investigation, and response on their own data foundation.

The label tells you little. The POC has to show what is actually doing the work.

Read this for the six checks that tell them apart: https://thehackernews.com/2026/07/how-to-evaluate-ai-soc-platform-in-2026.html

AI does not just make phishing cleaner. It changes the clock.

Guy Segal, CEO at Sygnia, frames the real gap: incident response plans were designed for slower, human-led attacks, while AI can compress recon, deception, payload changes, and the path to business impact.

See where your IR plan breaks when AI compresses the timeline: https://thehackernews.com/expert-insights/2026/07/ai-speed-attacks-are-forcing-rethink-of.html

🛑 A streaming box. A browser prompt. A fake repo. An AI agent reading the wrong thing.

This week’s cyber risks did not look exotic. That was the problem.

Monday Recap: proxy botnets, browser ransomware research, AI agent tricks, fake PoC malware, ClickFix attacks, and more.

Read the full recap: https://thehackernews.com/2026/07/monday-recap-proxy-botnets-browser.html

🔐 It’s easier to log in than hack in.

Attackers don’t need a zero-day if weak, reused, or compromised passwords are already sitting in Active Directory.

Specops’ Password Auditing Guide shows how to uncover hidden password risks, spot policy gaps, and start auditing AD with a free tool.

Download the guide to learn more: https://thn.news/audit-active-directory

🚨 CVE-2026-20896 saw its first in-the-wild attempt 13 days after disclosure.

The Gitea Docker flaw lets reachable containers trust spoofed X-WEBAUTH-USER headers when reverse proxy auth is enabled.

See which setups are exposed and where the probe stopped: https://thehackernews.com/2026/07/threat-actors-probe-gitea-docker-flaw.html

🔥 A new 16-year-old #Linux KVM flaw lets a rooted nested VM crash the x86 host and take down other tenants on the same machine.

Dubbed "Januscape" (CVE-2026-53359), the bug sits in KVM’s shadow MMU.

Researcher says a full guest-to-host escape exploit also exists in a controlled setup; only the host-crash PoC is public.

Explained here: https://thehackernews.com/2026/07/16-year-old-linux-kvm-flaw-lets-guest.html

🚨 Iran-linked Cavern Manticore is targeting Israeli government and IT organizations with 'Cavern,' a modular .NET C2 framework.

Researchers say the group abused RMM tools, then used DLL side-loading, NativeAOT modules, and AppDomain isolation to hide post-compromise activity.

How Cavern works: https://thehackernews.com/2026/07/iran-linked-hackers-use-new-cavern-c2.html

⚠️ BeyondTrust fixed four RS and PRA flaws. Two are critical pre-auth bugs that could let unauthenticated attackers bypass access controls, but only under specific auth configurations.

No exploitation is reported. RS 25.3.3 and PRA 25.3.3 carry the fixes.

Check which setups need patching: https://thehackernews.com/2026/07/beyondtrust-patches-critical-auth.html

🛑 WARNING - Several Tenda firmware builds embed an undocumented auth backdoor in /bin/httpd.

CVE-2026-11405 checks sys.rzadmin.password after normal login fails. If the supplied password matches, it creates an admin session.

Still unpatched.

Read: https://thehackernews.com/2026/07/certcc-warns-of-hidden-admin-backdoor.html

🚨 Suspected China-aligned UNK_MassTraction exploited now-patched Roundcube flaws against U.S. and Canadian university departments.

IceCube stole credentials, 2FA data, and cookies, then used CVE-2025-49113 to drop VShell or a web shell.

How the mail server compromise worked: https://thehackernews.com/2026/07/suspected-china-aligned-hackers-exploit.html

Breach transparency remains one of cybersecurity’s toughest governance challenges.

Bruce Sussman of Bitdefender highlights a key finding from the 2026 Cybersecurity Assessment: 55.2% of professionals who experienced a breach were told to keep it confidential—even when they believed it should be reported to authorities.

Why awareness still doesn’t translate into readiness: https://thehackernews.com/expert-insights/2026/07/breach-transparency-remains.html

🔥 Alleged Scattered Spider hacker caught by his own Windows device ID.

Per a new court filing, he hid the attack behind proxies and fake names.

But #Microsoft records tied one device ID to his ngrok access, then to his personal accounts and trips. That overlap is how the FBI found him.

Criminals, take notes 🠖 https://thehackernews.com/2026/07/court-filing-reveals-windows-device-id.html

AI has changed the software supply chain question.

It is no longer just which packages, versions, and dependencies made it into the code. It is which agents, MCP tools, models, and prompts shaped the build.

What supply chain security looks like with AI in scope: https://thehackernews.com/2026/07/what-changes-when-your-software-supply.html

Writer AI patched WriteOut, a one-click session isolation flaw in its enterprise AI platform.

A victim who clicked a public preview link while logged in could have their session cookie forwarded into an attacker-controlled sandbox, where it could be recovered and replayed.

How the preview proxy crossed tenant lines: https://thehackernews.com/2026/07/writer-ai-flaw-could-let-agent-previews.html

4 Cities. 4 Intimate roundtables. One urgent conversation. 🌍

Security has no shortage of conversations. Honest ones are harder to find.

This summer, TryHackMe is bringing senior security leaders together in-person for Post-Mythos: Watch the Incident Readiness Gap. Their first roundtable series across Europe and the US.

Built for leaders who want to debrief on how to close the gaps that actually matter. The series will cover:

🧠 Where human judgement sits when AI is running the attack.
🛡️ What it takes to build a proactive, AI-fluent SOC.
⚠️ Whether your team is ready for a threat class that’s already here.

Pick your city. Claim your seat. 👉🏼

📍San Francisco: https://thn.news/roundtableseries-sfo
📍Dallas: https://thn.news/roundtableseries-dallas
📍New York City: https://thn.news/roundtableseries-ny
📍 Paris: https://thn.news/roundtableseries-paris

⚡ New GitLost technique can trick #GitHub Agentic Workflows into leaking private repo data.

A public issue can make an agent read a private repo and post its contents in a public comment.

In one test, adding “Additionally” to the malicious instruction was enough to get past GitHub’s guardrail.

Learn how the leak works - https://thehackernews.com/2026/07/public-github-issue-could-trick-github.html