Researchers at Nebula Security disclosed GhostLock, a Linux kernel flaw that lets any logged-in user gain full root on an unpatched machine, and it also escapes containers. Tracked as CVE-2026-43499, the bug has shipped by default in essentially every mainstream distribution since 2011 and needs no special permissions or network access; ordinary threading calls from any local program are enough. Nebula built a working exploit it says is 97 percent reliable and published the code, though no in-the-wild use is known yet. There is no full workaround, patch availability is uneven across distributions, and one earlier fix introduced a separate crash bug, so confirm the fixed package version.
Ubiquiti has released fixes for 25 vulnerabilities across its UniFi ecosystem, including several critical flaws that let a network-based attacker fully compromise devices. The most severe, CVE-2026-50746 rated 10.0, is an improper access-control bug in the UniFi Connect application that allows unauthenticated command injection on the host. Others rated 9.9, spanning UniFi Talk, Access, Protect, and UniFi OS, include SQL injection, command injection, and server-side request forgery that lead to privilege escalation, and some can be chained to bypass authentication requirements. The flaws affect widely deployed gateways and consoles like the UDM family. Ubiquiti lists no interim workarounds, so updating each affected product to its fixed version is the only mitigation.
Microsoft has patched RoguePlanet, a Microsoft Defender flaw that a public proof-of-concept had been exploiting since June to gain SYSTEM privileges on fully updated Windows machines. Tracked as CVE-2026-50656, the bug is a race condition in Defender's scanning engine: a low-privileged attacker who already has code running on a host can win a timing window to spawn a command shell as SYSTEM, and the public exploit worked whether or not real-time protection was enabled. It does not provide initial access, but it turns any foothold into full local control, useful for disabling defenses, stealing credentials, and moving laterally. It is the fourth Defender flaw disclosed by the same researcher this year.
BeyondTrust has patched two critical flaws in its Remote Support and Privileged Remote Access products that let an unauthenticated, network-positioned attacker bypass authentication and reach the appliance, including accounts with elevated privileges. The bugs, CVE-2026-40138 and CVE-2026-40139, both rated 9.2, sit in the authentication subsystem and depend on a specific authentication configuration being enabled. Cloud-hosted customers were patched automatically in April, but self-hosted deployments on version 25.3.2 or earlier need to update themselves. BeyondTrust has not reported exploitation, but its remote-support products have a history of being attacked, including flaws used to breach the US Treasury and to deploy ransomware, so internet-facing appliances should be patched quickly.
CERT/CC has warned that several Tenda router firmware versions contain an undocumented authentication backdoor that grants full administrative access to the web management interface. Tracked as CVE-2026-11405, the flaw lives in the login function of the router's web server: if normal password checking fails, the firmware compares the supplied password against a hidden value stored in the device configuration and, on a match, grants admin access regardless of the username. It affects models including the FH1201, W15E, AC10, AC5, and AC6, is baked into the firmware, and cannot be disabled from the interface. Tenda has not responded, so there is no fix, and public exploit tooling is already scanning for vulnerable devices.
Varonis disclosed a now-patched flaw in Google Cloud's Dialogflow CX, the platform many companies use to build customer-service, financial, and healthcare chatbots. Because all chatbots using the platform's custom-code feature in one Google Cloud project shared a single execution environment with a writable setup file and no isolation, an attacker who could edit one agent, needing only a single low-level permission, could overwrite that file and hijack every chatbot in the project. From there they could read live conversations, steal shared data, and make bots ask for passwords. Google reported no exploitation before fixing it and no customer action is now required, but the case shows how AI features inherit cloud risks.
A use-after-free flaw in Linux's KVM hypervisor, nicknamed Januscape and hidden in the code for about 16 years, lets a virtual machine attack the physical host it runs on. Tracked as CVE-2026-53359, it sits in the shadow memory code that KVM uses on both Intel and AMD systems when nested virtualization is enabled. From inside a guest with root, an attacker can corrupt host kernel memory: the public proof-of-concept crashes the entire host, taking down every other tenant on that machine, and the researcher says a private exploit can run code as root on the host. The fix reached mainline Linux in June, and distributions are shipping updated kernels now.
Attackers have started probing a critical flaw in Gitea's official Docker images, the self-hosted Git service used by many development teams. The images shipped a configuration that trusted the X-WEBAUTH-USER header from any source address, so with reverse-proxy login enabled, anyone who could reach the port could send that header and be authenticated as any user, no password required. Tracked as CVE-2026-20896 and rated 9.8, it was fixed in version 1.26.3 late last month, which removes the wildcard and makes reverse-proxy authentication opt-in. Sysdig reported the first exploitation attempt 13 days after disclosure, so far just reconnaissance, against some of the roughly 6,200 internet-facing Gitea instances.
One of the critical ColdFusion vulnerabilities Adobe patched last week is now being exploited in the wild. The flaw, CVE-2026-48282, is a path-traversal issue rated 10.0 that lets an attacker run arbitrary code on a ColdFusion server, and it was among seven top-severity bugs Adobe fixed in ColdFusion 2025 and 2023. Adobe had flagged the update as high priority given ColdFusion's history as an attacker and ransomware target, and exploitation has followed quickly. Organizations that had not yet applied the update are now in an active-threat window, especially any ColdFusion servers reachable from the internet, which are the most exposed to opportunistic attacks.
Researchers found a flaw in Opera GX, the gaming-focused version of the Opera browser, that let a malicious website silently install a browser add-on, known as a mod, and use it to pull specific data from the pages the victim later visited. In a proof of concept, they reconstructed a signed-in user's full Gmail address from a single visit, with no clicks required. The issue stemmed from how the browser handled automatic mod installation, giving a hostile page more power than it should have. Opera has patched the flaw and says it found no evidence of exploitation before the fix, but users should update.