Upgrade Apache Roller to 6.1.6+ now; if you can't, set enabledForExtensions to false in web.xml or remove the XML-RPC servlet mapping and restart.
What it is AI
Apache Roller's XML-RPC endpoint deserializes attacker-supplied data before checking who sent it, letting an unauthenticated remote attacker run arbitrary code (RCE) on the blogging server.
Patch vCenter now to 9.1.1.0200, 9.0.2.0200, 8.0 U3m or 5.2.5 depending on your version; also patch Telco Cloud Platform/Infrastructure.
What it is AI
Broadcom's VMware vCenter has a flaw in its Directory Service that lets an unauthenticated attacker with network access to vCenter bypass login and gain unauthorized access to the system.
Apply SAP's security patch day fixes for NetWeaver Enterprise Portal now and review SAP Note 3660969 for required permissions.
What it is AI
SAP NetWeaver Enterprise Portal lets an unauthenticated attacker inject Java Naming and Directory Interface (JNDI) properties or a lookup URL, redirecting the server to an attacker-controlled JNDI provider and disclosing or altering information about the server.
Update REDCap to the vendor's fixed release now; CISA's Nov 21, 2026 remediation deadline has passed.
What it is AI
REDCap's public survey routing and data-import logic let an unauthenticated attacker reach an unintended internal controller and supply a crafted file path, running arbitrary code on the server if they know a valid survey's hash.
Migrate off end-of-life Axis 1.x to Axis2/Java, or apply Apache's patch commit 41f3b4e; no official 1.x fix is coming.
What it is AI
Apache Axis 1.x loads and instantiates any class named in an inbound SOAP fault, running its static initializer first. With a suitable class on the classpath this gives an unauthenticated attacker remote code execution that per-service authentication cannot block.
Patch LiteLLM now via the vendor or Red Hat updates; restrict network access to the guardrails test endpoint until patched.
What it is AI
LiteLLM, an open-source gateway for routing requests to AI model APIs, exposes a /guardrails/test_custom_code endpoint that lets a remote attacker rewrite Python bytecode and run arbitrary code (RCE) on the server, needing only low-level access to reach it.
Update DSM to 7.2-64570-4, 7.2.1-69057-6 or 7.2.2-72806-1 (BeeStation OS to 1.1-65374) now.
What it is AI
Synology DiskStation Manager (DSM)'s system plugin daemon fails to properly encode output, letting a remote, unauthenticated attacker run arbitrary code over unspecified network vectors.
A flaw in Puppet Server's report handling lets a client holding any agent certificate run arbitrary code as the server's account, via a crafted report. If the certificate authority (CA) auto-signs requests, an unauthenticated attacker can obtain such a certificate first.
Check Point Gaia DynamicCLI socket root command execution
What to do AI
Apply the mitigations in Check Point sk1000362 now; restrict local shell access to affected appliances and review accounts for unexpected root activity.
What it is AI
On Check Point gateways and management servers, the DynamicCLI daemon executes with root privileges and listens on a world-writable local socket without checking who connects. Any authenticated local user can send it a command that then executes as root.
UnrealIRCd labeled-response server command injection
What to do AI
Upgrade all linked UnrealIRCd servers to 6.2.7.1 or later now; review server-to-server traffic for commands injected by ordinary users.
What it is AI
UnrealIRCd's labeled-response module echoes a client-supplied message label into replies without escaping it. On networks with two or more linked servers, any connected user can smuggle server-to-server commands that the network then executes with server authority.
Update Chrome to 157.0.8103.50 or later now and restart the browser; update Edge, Opera and other Chromium-based browsers as their fixes ship.
What it is AI
A use-after-free memory flaw in Google Chrome before 157.0.8103.50 lets a crafted web page execute arbitrary code inside the browser sandbox. It is exploited in the wild, and Microsoft Edge, Opera and other Chromium-based browsers may also be affected.
Microsoft Outlook missing unsafe-action warning spoofing
What to do AI
Apply the remediation in Microsoft's advisory for this CVE now; brief users on spoofed messages until every client is updated.
What it is AI
Microsoft Office Outlook fails to warn the user before a potentially unsafe action, letting an unauthenticated remote attacker carry out spoofing over the network. Exploitation requires the targeted user to interact, and can expose sensitive information.
Browse by year or tag
Every card is also on a plain list: under the year CISA added it to KEV, its catalog of Known Exploited Vulnerabilities, and under each of its tags that several cards share.