← Vulnerabilities of the weekVulnerabilities of the week: November 9–15, 2026
The week’s top vulnerabilities: what they are and how they could harm you. Open an explanation to see who is affected and what to do.
#1 · CVE-2026-116779
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 9.8 · EPSS: 2.5% chance of an attack attempt in the next 30 days
What this is
The program freeswitch handles telecom work on a server. A specially made incoming message can make it treat part of that message as a command and run it.
How it can harm you
If the server accepts that message, an attacker can run their own program on the server with the service's privileges. The description also says this can expose configuration and credentials and let the attacker control calls; any authenticated profile account is sufficient, and some affected configurations can remove that barrier for stock directory users without password material.
Read the explanation →#2 · CVE-2026-123218
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 9.8 · EPSS: 0.5% chance of an attack attempt in the next 30 days
What this is
IBM Aspera Faspex has a flaw in user authorization. A specific message may skip normal permission checks and give full administrator access.
How it can harm you
The reports say a crafted message can bypass normal permissions and give full administrator rights in IBM Aspera Faspex. That could let an attacker use administrator-only functions. The available reports do not clearly state whether the attacker must already have a valid user account.
Read the explanation →#3 · CVE-2026-102895
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 9.8 · EPSS: 0.5% chance of an attack attempt in the next 30 days
What this is
gpustack is server software. A flaw in gpustack/api/auth.py can wrongly treat some network requests as the built-in administrator and let a person skip sign-in.
How it can harm you
If a remote request is accepted as if it came from 127.0.0.1 under the default setting, an outsider may get administrator access without signing in. From there they may change users and models, create a lasting administrator key, read worker and cloud-provider secrets, and run their own program on the server through /v2/inference-backends/from-yaml.
Read the explanation →#4 · CVE-2026-116606
attacks observed · public exploit on GitHub · Nuclei template · KEV: attacks observed in the wild · CVSS 9.3 · EPSS: 10.9% chance of an attack attempt in the next 30 days
What this is
Barracuda Email Security Gateway (ESG) has a broken sign-in check. Someone on the network can use its configuration import function to load a crafted, compatible backup archive without signing in.
How it can harm you
If an outsider can reach Barracuda Email Security Gateway (ESG) over the network and has a crafted, compatible backup archive, they can restore it without signing in and change the appliance's global configuration. No user action is needed.
Read the explanation →#5 · CVE-2026-125350
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 8.8 · EPSS: 1.0% chance of an attack attempt in the next 30 days
What this is
Google Chrome has a memory handling flaw. A specially made page can trigger it and make the browser use data that should no longer be used.
How it can harm you
If someone on the Internet can get a person to open a specially made page in Google Chrome, that attacker may run their own program inside the Google Chrome sandbox. No sign-in is needed, but the available facts do not say how an attacker could go beyond that restricted area.
Read the explanation →#6 · CVE-2026-111560
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 8.6 · EPSS: 9.1% chance of an attack attempt in the next 30 days
What this is
Certain Barracuda Networks appliance products have a problem in the timezone setting. A specially written timezone value can make the device run commands when that value is later handled.
How it can harm you
A person who can already sign in as an administrator over the network can place a crafted timezone value and, when the device later handles it, make it run their own commands as root. That can let them change settings or files and disrupt how the device works; the exact follow-on effects are not further detailed here.
Read the explanation →#7 · CVE-2026-114713
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 8.6 · EPSS: 0.5% chance of an attack attempt in the next 30 days
What this is
The labeled-response module in UnrealIRCd does not safely handle a user's label when the server echoes it in replies. This lets crafted text be interpreted as server-to-server commands.
How it can harm you
On a network with two or more linked servers, any user can inject commands that another server may treat as coming from a server. This can let them make changes with server authority; the exact results are not described in the available facts.
Read the explanation →#8 · CVE-2026-123039
attacks observed · public exploit on GitHub · KEV: attacks observed in the wild · CVSS 7.8 · EPSS: 2.2% chance of an attack attempt in the next 30 days
What this is
Quantum Security Gateway and Quantum Security Management include a local service that accepts commands through /tmp/dynamic_socket without checking which local program is allowed to connect. A signed-in local user can use this to make the service run a command as root.
How it can harm you
Someone who already has a local account on the server can send a command to that local socket and have it run as root, with no extra approval and no user action. That can let them run their own program on the server and then read private files or passwords, change or delete records, or stop services. Whether such a local account exists depends on your setup.
Read the explanation →#9 · CVE-2026-20320
attacks observed · KEV: attacks observed in the wild · CVSS 7.5 · EPSS: 1.0% chance of an attack attempt in the next 30 days
What this is
Cisco BroadWorks handles specially made input in an unsafe way. A crafted message can make the system reveal configuration information or files that should not be exposed.
How it can harm you
An outsider can try this over the network without signing in and without any action from you. They may be able to read configuration information or files that the Cisco BroadWorks user can read; the exact files are not stated here.
Read the explanation →
Check your server for free →